OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 3: XML Encoding Standard Approved OGC Standard • OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 3: XML Encoding Standard Peng Yue Editor Ruixiang Liu Editor Kaixuan Wang Editor Submission Date: 2026-03-01 Approval Date: 2026-04-28 Publication Date: 2026-08-20 External identifier of this OGC® document: http://www.opengis.net/doc/is/tdml-part3/1.0 Internal identifier of this OGC® document: 24-007r4 Version: 1.0.1 Additional Formats: PDF OGC Standard Approved Notice This document is an OGC Member approved international standard. This document is available on a royalty free, non-discriminatory basis. Recipients of this document are invited to submit, with their comments, notification of any relevant patent rights of which they are aware and to provide supporting documentation. Document number: 24-007r4 Document type: OGC Standard Document subtype: Encoding Document stage: Approved Document language: English Copyright notice Copyright © 2026 Open Geospatial Consortium To obtain additional rights of use, visit https://www.ogc.org/legal Note Attention is drawn to the possibility that some of the elements of this document may be the subject of patent rights. The Open Geospatial Consortium shall not be held responsible for identifying any or all such patent rights. Recipients of this document are requested to submit, with their comments, notification of any relevant patent claims or other intellectual property rights of which they may be aware that might be infringed by any implementation of the standard set forth in this document, and to provide supporting documentation. License Agreement Use of this document is subject to the license agreement at https://www.ogc.org/license Suggested additions, changes and comments on this document are welcome and encouraged. Such suggestions may be submitted using the online change request form on OGC web site: http://ogc.standardstracker.org/ Contents I . Abstract II . Keywords III . Preface IV . Security considerations V . Submitting Organizations VI . Submitters VII . Acknowledgements 1 . Scope 2 . Conformance 3 . Normative references 4 . Terms and definitions 5 . Conventions 5.1 . Identifiers 5.2 . Abbreviated Terms 6 . Overview 6.1 . Extensible Markup Language 7 . Requirements for TrainingDML-AI XML Encoding 7.1 . Requirements Class: Base 7.2 . Requirements Class: AI_TrainingDataset 7.3 . Requirements Class: AI_TrainingData 7.4 . Requirements Class: AI_Task 7.5 . Requirements Class: AI_Label 7.6 . Requirements Class: AI_Labeling 7.7 . Requirements Class: AI_DataQuality 7.8 . Requirements Class: AI_TDChangeset Annex A Abstract Test Suite A.1 . Introduction A.2 . Conformance Class: Base A.3 . Conformance Class: AI_TrainingDataset A.4 . Conformance Class: AI_TrainingData A.5 . Conformance Class: AI_Task A.6 . Conformance Class: AI_Label A.7 . Conformance Class: AI_Labeling A.8 . Conformance Class: AI_DataQuality A.9 . Conformance Class: AI_TDChangeset Annex B Example B.1 . TrainingDataset Encoding Examples B.2 . DataQuality Encoding Example B.3 . TDChangeset Encoding Example B.4 . Non-EO Imagery TrainingDataset Encoding Examples Annex C Revision History Bibliography Top I . Abstract The OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 3: XML Encoding Standard defines requirements for encoding AI training datasets as Extensible Markup Language (XML). XML is a widely utilized format for encoding data in various applications. It employs a markup structure where data is enclosed in tags represented by opening and closing elements. These elements can have attributes and may contain nested elements, allowing for a hierarchical representation of structured information. TrainingDML-AI Part 3 is based on the OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 1: Conceptual Model Standard. II . Keywords The following are keywords to be used by search engines and document catalogues. ogcdoc, OGC document, artificial intelligence, machine learning, deep learning, earth observation, remote sensing, training data, training sample, encoding, XML III . Preface Attention is drawn to the possibility that some of the elements of this document may be the subject of patent rights. The Open Geospatial Consortium shall not be held responsible for identifying any or all such patent rights. IV . Security considerations No security considerations have been made for this Standard. V . Submitting Organizations The following organizations submitted this Document to the Open Geospatial Consortium (OGC): Wuhan University Luoyao Information Technology Co., Ltd Pixalytics Ltd WiSC Enterprises George Mason University Laboratoire d'Informatique de Grenoble Wuhan University of Technology Hubei University Chongqing Changan Zhitu Technology Co., Ltd VI . Submitters All questions regarding this submission should be directed to the editors or the submitters: Name Affiliation Peng Yue Wuhan University Ruixiang Liu Wuhan University Kaixuan Wang Wuhan University Haoru Wu Wuhan University Boyi Shangguan Luoyao Information Technology Co., Ltd Samantha Lavender Pixalytics Ltd Jim Antonisse WiSC Enterprises Liping Di George Mason University Eugene Yu George Mason University Danielle Ziébelin Laboratoire d’Informatique de Grenoble Liangcun Jiang Wuhan University of Technology Lei Hu Hubei University Mingda Zhang Hubei University Kai Yan Chongqing Changan Zhitu Technology Co., Ltd VII . Acknowledgements Thanks to the members of the TrainingDML-AI Standards Working Group of the OGC as well as all contributors of change requests and comments. In particular: Scott Simmons, Carl Reed, Sam Meek, Yongyi Mi, Chenxiao Zhang, Zhipeng Cao, Shuaiqi Liu, Ming Zhao, Hanwen Xu, Haipeng Deng, Baoxin Teng. 1 . Scope This OGC TrainingDML — AI Part 3: XML Encoding Standard defines an XML encoding for the exchange of training datasets. The TrainingDML — AI Part 3 Standard provides an XML-based encoding for the exchange of information describing training datasets, both within and between different organizations. The document model is derived from the conceptual models defined in the OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 1: Conceptual Model Standard. 2 . Conformance This Standard defines an XML encoding for AI training datasets. The standardization target for this Standard is: TrainingDML-AI XML Encoding Schema Conformance with this Standard shall be checked using all the relevant tests specified in Annex A . The framework, concepts, and methodology for testing, and the criteria to be achieved to claim conformance are specified in the OGC Compliance Testing Policies and Procedures and the OGC Compliance Testing website . All requirements-classes and conformance-classes described in this document are owned by the standard identified. This standard identifies eleven (11) conformance classes. Each conformance class is defined by one requirements class. The tests in Annex A are organized by requirements class. So an implementation of each conformance class must pass all tests specified in Annex A for the respective requirements class. Of these eleven conformance classes, only the AI_TrainingDataset conformance class is mandatory. All other conformance classes are optional. In the case where a conformance class has a dependency on another conformance class, that conformance class should also be implemented. If AI_TrainingDataset conformance class is implemented, all other conformance classes will eventually be implemented based on the dependencies between the conformance classes. Therefore, all conformance classes can also be considered mandatory. 3 . Normative references The following documents are referred to in the text in such a way that some or all of their content constitutes requirements of this document. For dated references, only the edition cited applies. For undated references, the latest edition of the referenced document (including any amendments) applies. OGC: OGC 07-036, OpenGIS Geography Markup Language (GML) Encoding Standard, 2007 W3C Recommendation: W3C XML Schema Definition Language (XSD) 1.1 Part 1: Structures, 2012 W3C Recommendation: W3C XML Schema Definition Language (XSD) 1.1 Part 2: Datatypes, 2012 IETF: RFC 2046, Multipurpose Internet Mail Extensions (MIME) Part Two: Media Types, 1996 ISO 19101-1:2014 Geographic information — Reference model — Part 1: Fundamentals ISO 19103:2024 Geographic information — Conceptual schema language ISO 19115-1:2014 Geographic information — Metadata — Part 1: Fundamentals ISO 19156:2023 Geographic information — Observations, measurements and samples ISO 19157-1 Geographic information — Data quality — Part 1: General requirements 4 . Terms and definitions This document uses the terms defined in OGC Policy Directive 49 , which is based on the ISO/IEC Directives, Part 2, Rules for the structure and drafting of International Standards. In particular, the word “shall” (not “must”) is the verb form used to indicate a requirement to be strictly followed to conform to this document and OGC documents do not use the equivalent phrases in the ISO/IEC Directives, Part 2. This document also uses terms defined in the OGC Standard for Modular specifications ( OGC 08-131r3 ), also known as the ‘ModSpec’. The definitions of terms such as standard, specification, requirement, and conformance test are provided in the ModSpec. For the purposes of this document, the following additional terms and definitions apply. This document used the terms defined in OGC Policy Directive 49 , which is based on the ISO/IEC Directives, Part 2, Rules for the structure and drafting of International Standards. In particular, the word “shall” (not “must”) is the verb form used to indicate a requirement to be strictly followed to conform to this Standard and OGC documents do not use the equivalent phrases in the ISO/IEC Directives, Part 2. For the purposes of this document, the following additional terms and definitions apply. 4.1 . Artificial Intelligence (AI) refers to a set of methods and technologies that can empower machines or software to learn and perform tasks like humans. SOURCE: OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 1: Conceptual Model Standard 4.2 . Machine Learning (ML) is an important branch of artificial intelligence that gives computers the ability to improve their performance without explicitly being programmed to do so. ML processes create models from training data by using a set of learning algorithms, and then can use these models to make predictions. Depending on whether the training data include labels, the learning algorithms can be divided into supervised and unsupervised learning. SOURCE: OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 1: Conceptual Model Standard 4.3 . Deep Learning (DL) is a subset of machine learning, which is essentially a neural network with three or more layers. The number of layers is referred to as depth. While a neural network with a single layer can still make approximate predictions, additional hidden layers can help to optimize and refine for accuracy. SOURCE: https://www.ibm.com/topics/deep-learning 4.4 . Dataset identifiable collection of data [SOURCE: ISO 19115‑1:2014, 4.3] Note 1 to entry: A dataset can be a smaller grouping of data which, though limited by some constraint such as spatial extent or feature type, is located physically within a larger dataset. Theoretically, a dataset can be as small as a single feature or feature attribute contained within a larger dataset. A hardcopy map or chart can be considered a dataset. 4.5 . Training Dataset a collection of samples, often labelled with known terms or expected values for supervised learning. A training dataset can be divided into training, validation, and test sets. Training samples are different from samples in OGC Observations & Measurements (O&M). They are often collected in purposive ways that deviate from purely probability sampling, with known or expected results labeled as values of a dependent variable for generating a trained predictive model. SOURCE: OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 1: Conceptual Model Standard Note 1 to entry: A training sample label is different from those on a geographical map, which are known as map labels or annotations. 4.6 . Label refers to known or expected results annotated as values of a dependent variable in training samples. A training sample label is different from those on a geographical map, which are known as map labels or annotations. SOURCE: OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 1: Conceptual Model Standard 4.7 . Class <classification> result of a classification process as part of a classification system which subdivides concepts within a given topic area. [SOURCE: ISO 19144-2:2023, 3.1.6] 4.8 . Task the specific goal that an AI application want to achieve. SOURCE: OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 1: Conceptual Model Standard 4.9 . Provenance information about entities, activities, and people involved in producing a piece of data or thing, which can be used to form assessments about its quality, reliability or trustworthiness. In this standard provenance is a record of how training data were prepared. SOURCE: W3C ( https://www.w3.org/TR/prov-overview/ ) 4.10 . Quality degree to which a set of inherent characteristics of an object fulfils requirements [ISO 9000:2015, 3.6.2, modified－Notes 1 and 2 to entry have been deleted]. Quality of training data (such as data imbalance and mislabeling) can impact the performance of AI/ML models. SOURCE: OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 1: Conceptual Model Standard 4.11 . Earth Observation data and information collected about our planet, whether atmospheric, oceanic or terrestrial. This includes space-based or remotely-sensed data, as well as ground-based or in situ data. SOURCE: GEO ( https://earthobservations.org/geo_wwd.php ) 4.12 . Scene Classification task of identifying scene categories of images, on the basis of a training set of images whose scene categories are known. SOURCE: OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 1: Conceptual Model Standard 4.13 . Object Detection task of recognizing objects such as cars from images. The objects are often localized using bounding boxes. SOURCE: OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 1: Conceptual Model Standard 4.14 . Semantic Segmentation task of assigning class labels to pixels of images or points of point clouds. SOURCE: OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 1: Conceptual Model Standard 4.15 . Change Detection recognition of changes between images acquired at different times. SOURCE: OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 1: Conceptual Model Standard 4.16 . 3D Model Reconstruction task that builds 3D objects and scenes from multi-view images. SOURCE: OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 1: Conceptual Model Standard 4.17 . Generative Model is one of the methods of large model training, which improve model performance through unsupervised pre-training. In the fine-tuning phase, labeled data plays a critical role in optimizing the model for specific vertical domains or tasks. By incorporating labeled data, the model can learn to accurately identify and extract relevant features, leading to better performance on specific downstream tasks. Overall, the combination of generative models and fine-tuning with labeled data can significantly improve the performance of large models in specialized domains or tasks. SOURCE: OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 1: Conceptual Model Standard 4.18 . Extensible Markup Language (XML) is a simple, very flexible text format derived from SGML (ISO 8879). Originally designed to meet the challenges of large-scale electronic publishing, XML is also playing an increasingly important role in the exchange of a wide variety of data on the Web and elsewhere. SOURCE: https://www.w3.org/XML/ 4.19 . XML Schema is a means for defining the structure, content and semantics of XML documents. SOURCE: https://www.w3.org/XML/Schema 4.20 . Training Dataset Publisher refers to the entity or individual responsible for creating and releasing the XML-based serialization syntax for geospatial training datasets, as defined in the TrainingDML-AI Part 3: XML Encoding Standard. 5 . Conventions This section provides details and examples for any conventions used in the document. 5.1 . Identifiers The normative provisions in this Standard are denoted by the URI: http://www.opengis.net/spec/TrainingDML-AI-3/1.0 All requirements and conformance tests that appear in this document are denoted by partial URIs which are relative to this base. 5.2 . Abbreviated Terms In this document, the following abbreviations and acronyms are used or introduced: AI Artificial Intelligence DL Deep Learning EO Earth Observation IETF Internet Engineering Task Force ISO International Organization for Standardization JSON JavaScript Object Notation ML Machine Learning OGC Open Geospatial Consortium RS Remote Sensing TD Training Data UML Unified Modelling Language URL Uniform Resource Locator UTC Coordinated Universal Time W3C World Wide Web Consortium XML Extensible Markup Language 6 . Overview The TrainingDML — AI Part 3: XML Encoding Standard defines an XML-based serialization syntax for geospatial training datasets. While other serialization formats are possible, such alternatives are not discussed in this Standard. XML employs a formal class model with elements and attributes, allowing for a more structured representation of data. XML documents are hierarchical in nature, composed of nested elements that can contain both data and metadata. A training dataset document conforming to this Standard is an XML document whose root element is an “AI_TrainingDataset” element. 6.1 . Extensible Markup Language Extensible Markup Language (XML) is a versatile and widely used markup language designed for encoding documents in a format that is both human-readable and machine-readable. It provides a set of rules for defining custom markup languages and is often employed for representing structured data in a hierarchical and platform-independent manner. XML documents consist of elements, attributes, and text content organized within a tree-like structure. Elements are enclosed within tags, and attributes provide additional information about the elements. XML has widespread applicability and is utilized in various domains, including web development, data interchange, configuration files, and more. Its flexibility allows users to define their own document structures, making it suitable for a broad range of applications. XML is supported by many programming languages and technologies, contributing to its interoperability and adoption. When serialized, absent properties in XML are typically represented by leaving the corresponding elements empty, without any content. Alternatively, certain XML-based formats may use attributes to signify the absence of a property. These representations convey semantic equivalence. An empty or missing element signifies that no value has been assigned, distinct from the interpretation that the provided value is empty or nil. 7 . Requirements for TrainingDML-AI XML Encoding 7.1 . Requirements Class: Base 7.1.1 . Requirements Class: XML Base Type The XML Base Type requirements class defines the base requirements for XML encodings, which includes definitions of common types used in the TrainingDML-AI XML encoding. Requirements class /req/base/xmlbasetype Dependency XML Requirement 1 /req/base/xmlbasetype/xml Requirement 2 /req/base/xmlbasetype/datetime Requirement 3 /req/base/xmlbasetype/namedvalue Requirement 4 /req/base/xmlbasetype/url The first requirement is that a TrainingDML-AI XML document is a valid XML document. Requirement 1 /req/base/xmlbasetype/xml An instance SHALL be a conformant XML document, as defined by W3C Recommendation XML has a limited range of built-in types ( https://www.w3.org/TR/xmlschema11-2/ ). The following requirements provide standard XML representations of additional types required across all requirements within this specification. The DateTime is encoded as an XML element defined as one of three elements of type “date”, “time”, “dateTime”. Requirement 2 /req/base/xmlbasetype/datetime Each DateTime value SHALL be encoded as a text string defined in Section 3.3.7 , Section 3.3.8 or Section 3.3.9 of W3C XML Schema Definition Language (XSD) 1.1 Part 2: Datatypes. The specification of date and time in any XML encoding of training set data SHALL be specified in UTC. Examples: a) "2002-05-30T09:30:10Z" b) "2002-09-24" The NamedValue is encoded as an XML element with two elements named “key” and “value”. Typically, the “key” element represents a unique identifier or name for the value being described, while the “value” element contains the actual data associated with that identifier. The “key” element is an open field allowing for arbitrary keys to be used. The “value” element may be any types. Requirement 3 /req/base/xmlbasetype/namedvalue Each NamedValue value SHALL be encoded as an XML element with elements “key” and “value”, while the type of “key” is CharacterString. Examples: a) <key> forest </key> <value> RGB(0,255,255) </value> b) <key> precision </key> <value> 0.8 </value> The URL is encoded as an XML element defined as the type of “anyURI”. Requirement 4 /req/base/xmlbasetype/url Each URL value SHALL be encoded as a text string defined in Section 3.3.17 of W3C XML Schema Definition Language (XSD) 1.1 Part 2: Datatypes. Examples: a) "http://www.opengeospatial.org" b) "/file.txt" c) "./dataset/file.txt" 7.1.2 . Requirements Class: ISO Metadata Type The ISO Metadata Type requirements class defines the requirements for XML encoding of ISO metadata types. Requirements class /req/base/isometadatatype Dependency XML Requirement 5 /req/base/isometadatatype/band Requirement 6 /req/base/isometadatatype/extent Requirement 7 /req/base/isometadatatype/citation Requirement 8 /req/base/isometadatatype/scope Requirement 9 /req/base/isometadatatype/referencesystem Requirement 10 /req/base/isometadatatype/lineage Requirement 11 /req/base/isometadatatype/resolution The MD_Band is encoded as an XML element. Requirement 5 /req/base/isometadatatype/band Each MD_Band value SHALL be encoded as an XML element matching the XML Schema type as defined in: https://schemas.isotc211.org/19115/-1/mrc/1.3.0/content.xsd Examples: a) <mrc:name> <mcc:MD_Identifier> <mcc:code> <gco:CharacterString> red </gco:CharacterString> </mcc:code> </mcc:MD_Identifier> </mrc:name> b) <mrc:name> <mcc:MD_Identifier> <mcc:code> <gco:CharacterString> B4 </gco:CharacterString> </mcc:code> </mcc:MD_Identifier> </mrc:name> c) <mrc:boundMax> <gco:Real> 690 </gco:Real> </mrc:boundMax> <mrc:boundMin> <gco:Real> 630 </gco:Real> </mrc:boundMin> <mrc:boundUnits> <gml:UnitDefinition gml:id= "ID" > <gml:identifier codeSpace= "http://www.example.com/" > nm </gml:identifier> </gml:UnitDefinition> </mrc:boundUnits> The EX_Extent is encoded as an XML element. Requirement 6 /req/base/isometadatatype/extent Each EX_Extent value SHALL be encoded as an XML element matching the XML Schema type as defined in: https://schemas.isotc211.org/19115/-1/gex/1.3.0/extent.xsd Examples: a) <gex:geographicElement> <gex:EX_BoundingPolygon> <gex:polygon> <gml:Polygon> <gml:exterior> <gml:LinearRing> <gml:posList> 51.556272 -0.2803943 51.5562758 -0.2787397 51.5556539 -0.278736 51.5556501 -0.2803906 51.556272 -0.2803943 </gml:posList> </gml:LinearRing> </gml:exterior> </gml:Polygon> </gex:polygon> </gex:EX_BoundingPolygon> </gex:geographicElement> b) <gex:geographicElement> <gex:EX_GeographicBoundingBox> <gex:westBoundLongitude> <gco:Decimal> -171.76409 </gco:Decimal> </gex:westBoundLongitude> <gex:eastBoundLongitude> <gco:Decimal> -157.86768 </gco:Decimal> </gex:eastBoundLongitude> <gex:southBoundLatitude> <gco:Decimal> -14.42443 </gco:Decimal> </gex:southBoundLatitude> <gex:northBoundLatitude> <gco:Decimal> 21.31573 </gco:Decimal> </gex:northBoundLatitude> </gex:EX_GeographicBoundingBox> </gex:geographicElement> The CI_Citation is encoded as an XML element. Requirement 7 /req/base/isometadatatype/citation Each CI_Citation value SHALL be encoded as an XML element matching the XML Schema type as defined in: https://schemas.isotc211.org/19115/-1/cit/1.3.0/citation.xsd Example: <cit:title> <gco:CharacterString> Open Geospatial Consortium </gco:CharacterString> </cit:title> <cit:alternateTitle> <gco:CharacterString> OGC </gco:CharacterString> </cit:alternateTitle> <cit:identifier> <mcc:MD_Identifier> <mcc:code> <gco:CharacterString> artifact_id=104605 </gco:CharacterString> </mcc:code> <mcc:version> <gco:CharacterString> 1 </gco:CharacterString> </mcc:version> </mcc:MD_Identifier> </cit:identifier> The MD_Scope is encoded as an XML element. Requirement 8 /req/base/isometadatatype/scope Each MD_Scope value SHALL be encoded as an XML element matching the XML Schema type as defined in: https://schemas.isotc211.org/19115/-1/mcc/1.3.0/commonClasses.xsd Examples: a) <mcc:level> <mcc:MD_ScopeCode codeList= "http://standards.iso.org/iso/19115/resources/Codelist/cat/codelists.xml#MD_ScopeCode" codeListValue= "dataset" /> </mcc:level> <mcc:levelDescription> <mcc:MD_ScopeDescription> <mcc:dataset> <gco:CharacterString> whu_rs19 </gco:CharacterString> </mcc:dataset> </mcc:MD_ScopeDescription> </mcc:levelDescription> b) <mcc:level> <mcc:MD_ScopeCode codeList= "http://standards.iso.org/iso/19115/resources/Codelist/cat/codelists.xml#MD_ScopeCode" codeListValue= "attribute" /> </mcc:level> <mcc:levelDescription> <mcc:MD_ScopeDescription> <mcc:attributes> <gco:CharacterString> completeness </gco:CharacterString> </mcc:attributes> </mcc:MD_ScopeDescription> </mcc:levelDescription> c) <mcc:level> <mcc:MD_ScopeCode codeList= "http://standards.iso.org/iso/19115/resources/Codelist/cat/codelists.xml#MD_ScopeCode" codeListValue= "coverage" /> </mcc:level> <mcc:extent> <gex:EX_Extent> <gex:geographicElement> <gex:EX_GeographicBoundingBox> <gex:westBoundLongitude> <gco:Decimal> -171.76409 </gco:Decimal> </gex:westBoundLongitude> <gex:eastBoundLongitude> <gco:Decimal> -157.86768 </gco:Decimal> </gex:eastBoundLongitude> <gex:southBoundLatitude> <gco:Decimal> -14.42443 </gco:Decimal> </gex:southBoundLatitude> <gex:northBoundLatitude> <gco:Decimal> 21.31573 </gco:Decimal> </gex:northBoundLatitude> </gex:EX_GeographicBoundingBox> </gex:geographicElement> </gex:EX_Extent> </mcc:extent> d) <mcc:level> <mcc:MD_ScopeCode codeList= "http://standards.iso.org/iso/19115/resources/Codelist/cat/codelists.xml#MD_ScopeCode" codeListValue= "feature" /> </mcc:level> <mcc:extent> <gex:EX_Extent> <gex:geographicElement> <gex:EX_BoundingPolygon> <gex:polygon> <gml:Polygon> <gml:exterior> <gml:LinearRing> <gml:posList> 51.556272 -0.2803943 51.5562758 -0.2787397 51.5556539 -0.278736 51.5556501 -0.2803906 51.556272 -0.2803943 </gml:posList> </gml:LinearRing> </gml:exterior> </gml:Polygon> </gex:polygon> </gex:EX_BoundingPolygon> </gex:geographicElement> </gex:EX_Extent> </mcc:extent> The MD_ReferenceSystem is encoded as an XML object. Requirement 9 /req/base/isometadatatype/referencesystem Each MD_ReferenceSystem value SHALL be encoded as an XML object matching the XML Schema type as defined in: https://standards.iso.org/iso/19115/-3/mrs/1.0/referenceSystem.xsd Examples: a) <mrs:referenceSystemIdentifier> <mcc:MD_Identifier> <mcc:code> <gco:CharacterString> EPSG:3857 </gco:CharacterString> </mcc:code> </mcc:MD_Identifier> </mrs:referenceSystemIdentifier> b) <mrs:referenceSystemIdentifier> <mcc:MD_Identifier> <mcc:code> <gco:CharacterString> EPSG:4326 </gco:CharacterString> </mcc:code> </mcc:MD_Identifier> </mrs:referenceSystemIdentifier> The LI_Lineage is encoded as an XML object. Requirement 10 /req/base/isometadatatype/lineage Each LI_Lineage value SHALL be encoded as an XML object matching the XML Schema type as defined in: https://schemas.isotc211.org/19115/-1/mrl/1.3.0/lineage.xsd Examples: <mrl:LI_Lineage> <mrl:statement> <gco:CharacterString> The dataset was generated from high-resolution satellite imagery through a standardized preprocessing and manual labeling workflow for AI model training. </gco:CharacterString> </mrl:statement> <mrl:source> <mrl:LI_Source> <mrl:description> <gco:CharacterString> High-resolution multispectral satellite imagery acquired in 2023 covering urban areas. </gco:CharacterString> </mrl:description> <mrl:sourceCitation> <cit:CI_Citation> <cit:title> <gco:CharacterString> Urban Satellite Imagery Dataset 2023 </gco:CharacterString> </cit:title> </cit:CI_Citation> </mrl:sourceCitation> </mrl:LI_Source> </mrl:source> <mrl:processStep> <mrl:LI_ProcessStep> <mrl:description> <gco:CharacterString> Radiometric calibration and geometric correction were applied to the raw satellite images. </gco:CharacterString> </mrl:description> <mrl:rationale> <gco:CharacterString> To ensure spatial alignment and consistent reflectance values for subsequent labeling. </gco:CharacterString> </mrl:rationale> <mrl:stepDateTime> <gml:TimeInstant> <gml:timePosition> 2023-06-15T10:00:00Z </gml:timePosition> </gml:TimeInstant> </mrl:stepDateTime> </mrl:LI_ProcessStep> </mrl:processStep> <mrl:processStep> <mrl:LI_ProcessStep> <mrl:description> <gco:CharacterString> Trained annotators manually labeled buildings and roads using a GIS-based annotation platform. </gco:CharacterString> </mrl:description> <mrl:rationale> <gco:CharacterString> To generate high-quality ground truth data for supervised deep learning models. </gco:CharacterString> </mrl:rationale> <mrl:stepDateTime> <gml:TimePeriod> <gml:beginPosition> 2023-06-20 </gml:beginPosition> <gml:endPosition> 2023-07-10 </gml:endPosition> </gml:TimePeriod> </mrl:stepDateTime> </mrl:LI_ProcessStep> </mrl:processStep> </mrl:LI_Lineage> The MD_Resolution is encoded as an XML object. Requirement 11 /req/base/isometadatatype/resolution Each MD_Resolution value SHALL be encoded as an XML object matching the XML Schema type as defined in: https://standards.iso.org/iso/19115/-3/mri/1.0/identification.xsd Examples: a) <mri:equivalentScale> <mri:MD_RepresentativeFraction> <mri:denominator> <gco:Integer> 10000 </gco:Integer> </mri:denominator> </mri:MD_RepresentativeFraction> </mri:equivalentScale> b) <mri:distance> <gco:Distance uom= "m" > 10 </gco:Distance> </mri:distance> c) <mri:vertical> <gco:Distance uom= "m" > 1 </gco:Distance> </mri:vertical> d) <mri:angularDistance> <gco:Angle uom= "degree" > 0.01 </gco:Angle> </mri:angularDistance> e) <mri:levelOfDetail> <gco:CharacterString> High resolution imagery suitable for urban analysis </gco:CharacterString> </mri:levelOfDetail> 7.1.3 . Requirements Class: ISO Quality Type The ISO Quality Type requirements class defines the requirements for XML encoding of ISO quality types. Requirements class /req/base/isoqualitytype Dependency XML Requirement 12 /req/base/isoqualitytype/element The QualityElement object is encoded as an XML element with properties shown in Table 1 . Requirement 12 /req/base/isoqualitytype/element Each QualityElement value SHALL be encoded as an XML element with properties shown in Table 1 . Table 1 — QualityElement properties XML Property Definition Data type and values Obligation type The type of the quality element object. CharacterString [1..1] Mandatory measure Reference to measure used. MeasureReference [1..1] Mandatory evaluationMethod Evaluation information. EvaluationMethod [1..*] Mandatory result Value obtained from applying a data quality measure. QualityResult [1..*] Mandatory Example: <type> FormatConsistency </type> <measure> <measureDescription> Percentage of training samples with inconsistent image format </measureDescription> </measure> <evaluationMethod> <evaluationMethodDescription> Full test method to calculate the percentage of training samples with an inconsistent format </evaluationMethodDescription> </evaluationMethod> <result> <quantitativeResult> <value> 0 </value> <valueUnit> % </valueUnit> </quantitativeResult> </result> 7.1.4 . Requirements Class: Geospatial Type The Geospatial Type requirements class defines the requirements for XML encoding of geospatial types. Requirements class /req/base/geospatialtype Dependency XML Dependency GML Requirement 13 /req/base/geospatialtype/feature The encoding of one or more features follows GML for encoding a Feature element, with members “type”, “geometry” and “properties”. A Feature element represents a spatially bounded thing. Every Feature element is a GML element no matter where it occurs in a GML text. Requirement 13 /req/base/geospatialtype/feature Each Feature value SHALL be encoded using the GML feature encoding defined by OpenGIS in GML Section 3.2: http://www.opengis.net/gml/3.2 Examples of Feature encodings are: a) <gml:metaDataProperty> <gml:GenericMetaData> <truncated> 0.0 </truncated> <occluded> 0 </occluded> <alpha> -0.2 </alpha> </gml:GenericMetaData> </gml:metaDataProperty> <gml:location> <gml:GeometricComplex> <gml:element> <gml:Polygon> <gml:exterior> <gml:LinearRing> <gml:posList> 51.556272 -0.2803943 51.5562758 -0.2787397 51.5556539 -0.278736 51.5556501 -0.2803906 51.556272 -0.2803943 </gml:posList> </gml:LinearRing> </gml:exterior> </gml:Polygon> </gml:element> </gml:GeometricComplex> </gml:location> b) <gml:metaDataProperty> <gml:GenericMetaData> <iscrowd> 0 </iscrowd> <area> 2580 </area> </gml:GenericMetaData> </gml:metaDataProperty> <gml:location> <gml:GeometricComplex> <gml:element> <gml:LineString> <gml:posList> 1.0 1.0 2.0 2.0 </gml:posList> </gml:LineString> </gml:element> </gml:GeometricComplex> </gml:location> c) <gml:location> <gml:GeometricComplex> <gml:element> <gml:Point> <gml:pos> 1.0 1.0 </gml:pos> </gml:Point> </gml:element> </gml:GeometricComplex> </gml:location> 7.2 . Requirements Class: AI_TrainingDataset The AI_TrainingDataset requirements class defines an XML encoding for the AI_TrainingDataset module, which is based on the UML model specified in the TrainingDML-AI Part 1: Conceptual Model Standard. Requirements class /req/aitrainingdataset Dependency XML Dependency /req/base/xmlbasetype Dependency /req/base/isometadatatype Dependency /req/aitrainingdata Dependency /req/aitask Dependency /req/ailabeling Dependency /req/aidataquality Dependency /req/aitdchangeset Requirement 14 /req/aitrainingdataset/trainingdataset Requirement 15 /req/aitrainingdataset/metricsinliterature Requirement 16 /req/aitrainingdataset/eotrainingdataset The AI_TrainingDataset object is encoded as an XML element with properties shown in Table 2 . Requirement 14 /req/aitrainingdataset/trainingdataset Each AI_TrainingDataset object SHALL implement the Mandatory properties shown in Table 2 . Table 2 — AI_TrainingDataset properties XML Property Definition Data type and values Obligation type Type of the training dataset. “AI_AbstractTrainingDataset” Mandatory id Identification of the AI training dataset. CharacterString [1..1] Mandatory doi Digital object identifier of the AI training dataset. CharacterString [0..1] Optional scope Description of the scope of the training dataset. MD_Scope [0..1] Optional name Name of the AI training dataset. CharacterString [1..1] Mandatory description Description of the AI training dataset. CharacterString [1..1] Mandatory version Version number of the AI training dataset. CharacterString [0..1] Optional amountOfTrainingData Total number of training samples in the AI training dataset. Int [0..1] Optional createdTime Time when the AI training dataset was created. DateTime [0..1] Optional updatedTime Time when the AI training dataset was updated. DateTime [0..1] Optional license License description of the AI training dataset. CharacterString [1..1] Mandatory providers People or organizations who provide the AI training dataset. CharacterString [0..*] Optional keywords Keywords of the AI training dataset. CharacterString [0..*] Optional metricsInLIT Results of performance metrics achieved by AI/ML algorithms in the peer-reviewed literature. AI_MetricsInLiterature [0..*] Optional statisticsInfo Statistical results for training samples in each class. NamedValue [0..*] Optional dataSources Citation for the data sources. CI_Citation [0..*] Optional numberOfClasses Total number of classes in the AI training dataset. Int [0..1] Optional classificationScheme Classification scheme for classes used in the AI training dataset. CharacterString [0..1] Optional classes Classes used in the AI training dataset. NamedValue [1..1] Optional extent Spatial extent of the training dataset. EX_Extent [0..1] Optional referenceSystem Reference system of the training dataset. MD_ReferenceSystem [0..1] Optional lineage Information about the provenance, source(s), and/or the production process(es) applied to the training dataset. LI_Lineage [0..*] Optional tasks Task description of the training dataset. AI_Task [1..*] Mandatory labeling Provenance information of how the training dataset is labeled. AI_Labeling [0..*] Optional quality Quality information of the training dataset. DataQuality [0..*] Optional changesets Changeset between two versions of the training dataset. AI_TDChangeset [0..*] Optional data Training data in the training dataset. AI_AbstractTrainingData [1..*] Mandatory Example: <type> AI_AbstractTrainingDataset </type> <id> whu_rs19 </id> <name> WHU-RS19 </name> <description> Wuhan University-Remote Sensing 19 Categories (WHU-RS19) has 19 classes of remote sensing images scenes obtained from Google Earth </description> <license> CC BY-SA 4.0 </license> <amountOfTrainingData> 1013 </amountOfTrainingData> <createdTime> 2010-01-01T00:00:00 </createdTime> <providers> Wuhan University </providers> <keywords> Remote Sensing </keywords> <keywords> Scene Classification </keywords> <numberOfClasses> 19 </numberOfClasses> <classes> <key> Airport </key> <value> null </value> </classes> <classes> <key> Beach </key> <value> null </value> </classes> <classes> <key> Bridge </key> <value> null </value> </classes> <classes> <key> Commercial </key> <value> null </value> </classes> <classes> <key> Desert </key> <value> null </value> </classes> <classes> <key> Farmland </key> <value> null </value> </classes> <classes> <key> footballField </key> <value> null </value> </classes> <classes> <key> Forest </key> <value> null </value> </classes> <classes> <key> Industrial </key> <value> null </value> </classes> <classes> <key> Meadow </key> <value> null </value> </classes> <classes> <key> Mountain </key> <value> null </value> </classes> <classes> <key> Park </key> <value> null </value> </classes> <classes> <key> Parking </key> <value> null </value> </classes> <classes> <key> Pond </key> <value> null </value> </classes> <classes> <key> Port </key> <value> null </value> </classes> <classes> <key> railwayStation </key> <value> null </value> </classes> <classes> <key> Residential </key> <value> null </value> </classes> <classes> <key> River </key> <value> null </value> </classes> <classes> <key> Viaduct </key> <value> null </value> </classes> <extent> <gex:geographicElement> <gex:EX_GeographicBoundingBox> <gex:westBoundLongitude> <gco:Decimal> -180 </gco:Decimal> </gex:westBoundLongitude> <gex:eastBoundLongitude> <gco:Decimal> 180 </gco:Decimal> </gex:eastBoundLongitude> <gex:southBoundLatitude> <gco:Decimal> -90 </gco:Decimal> </gex:southBoundLatitude> <gex:northBoundLatitude> <gco:Decimal> 90 </gco:Decimal> </gex:northBoundLatitude> </gex:EX_GeographicBoundingBox> </gex:geographicElement> </extent> <mrs:MD_ReferenceSystem> <mrs:referenceSystemIdentifier> <mcc:MD_Identifier> <mcc:code> <gco:CharacterString> EPSG:4326 </gco:CharacterString> </mcc:code> </mcc:MD_Identifier> </mrs:referenceSystemIdentifier> </mrs:MD_ReferenceSystem> <tasks> <type> AI_EOTask </type> <id> whu_rs19-task </id> <description> Structural high-resolution satellite image indexing </description> <taskType> Scene Classification </taskType> </tasks> <data> <type> AI_EOTrainingData </type> <id> airport_01 </id> <dataSources> <cit:title> <gco:CharacterString> googleEarth </gco:CharacterString> </cit:title> </dataSources> <dataURL> image/Airport/airport_01.jpg </dataURL> <labels> <type> AI_SceneLabel </type> <class> Airport </class> </labels> </data> If the optional element AI_MetricsInLiterature is specified, this element is encoded as XML element with properties as shown in Table 3 . Requirement 15 /req/aitrainingdataset/metricsinliterature Each AI_MetricsInLiterature value SHALL implement the Mandatory properties shown in Table 3 . Table 3 — AI_MetricsInLiterature properties XML Property Definition Data type and values Obligation doi Digital object identifier of the peer-reviewed literature. CharacterString [1..1] Mandatory algorithm AI/ML algorithms used in the peer-reviewed literature. CharacterString [0..1] Optional metrics Metrics and results of AI/ML algorithms in the peer-reviewed literature. NamedValue [1..*] Mandatory Example: <doi> 10.1109/TGRS.2019.2917161 </doi> <algorithm> FACNN </algorithm> <metrics> <key> Overall Accuracy </key> <value> 0.9881 </value> </metrics> The AI_EOTrainingDataset object is encoded as an XML element with properties shown in Table 2 and Table 4 . Requirement 16 /req/aitrainingdataset/eotrainingdataset Each AI_EOTrainingDataset object SHALL implement the Mandatory properties both shown in Table 2 and Table 4 . Table 4 — AI_EOTrainingDataset properties XML Property Definition Data type and values Obligation type Type of the training dataset. “AI_EOTrainingDataset” Mandatory bands Description of the image bands used in the EO training dataset. MD_Band [0..*] Optional imageSize Size of the images used in the EO training dataset. CharacterString [0..1] Optional Example: <type> AI_EOTrainingDataset </type> <id> whu_rs19 </id> <name> WHU-RS19 </name> <description> Wuhan University-Remote Sensing 19 Categories (WHU-RS19) has 19 classes of remote sensing images scenes obtained from Google Earth </description> <license> CC BY-SA 4.0 </license> <amountOfTrainingData> 1013 </amountOfTrainingData> <createdTime> 2010-01-01T00:00:00 </createdTime> <providers> Wuhan University </providers> <keywords> Remote Sensing </keywords> <keywords> Scene Classification </keywords> <numberOfClasses> 19 </numberOfClasses> <classes> <key> Airport </key> <value> null </value> </classes> <classes> <key> Beach </key> <value> null </value> </classes> <classes> <key> Bridge </key> <value> null </value> </classes> <classes> <key> Commercial </key> <value> null </value> </classes> <classes> <key> Desert </key> <value> null </value> </classes> <classes> <key> Farmland </key> <value> null </value> </classes> <classes> <key> footballField </key> <value> null </value> </classes> <classes> <key> Forest </key> <value> null </value> </classes> <classes> <key> Industrial </key> <value> null </value> </classes> <classes> <key> Meadow </key> <value> null </value> </classes> <classes> <key> Mountain </key> <value> null </value> </classes> <classes> <key> Park </key> <value> null </value> </classes> <classes> <key> Parking </key> <value> null </value> </classes> <classes> <key> Pond </key> <value> null </value> </classes> <classes> <key> Port </key> <value> null </value> </classes> <classes> <key> railwayStation </key> <value> null </value> </classes> <classes> <key> Residential </key> <value> null </value> </classes> <classes> <key> River </key> <value> null </value> </classes> <classes> <key> Viaduct </key> <value> null </value> </classes> <extent> <gex:geographicElement> <gex:EX_GeographicBoundingBox> <gex:westBoundLongitude> <gco:Decimal> -180 </gco:Decimal> </gex:westBoundLongitude> <gex:eastBoundLongitude> <gco:Decimal> 180 </gco:Decimal> </gex:eastBoundLongitude> <gex:southBoundLatitude> <gco:Decimal> -90 </gco:Decimal> </gex:southBoundLatitude> <gex:northBoundLatitude> <gco:Decimal> 90 </gco:Decimal> </gex:northBoundLatitude> </gex:EX_GeographicBoundingBox> </gex:geographicElement> </extent> <mrs:MD_ReferenceSystem> <mrs:referenceSystemIdentifier> <mcc:MD_Identifier> <mcc:code> <gco:CharacterString> EPSG:4326 </gco:CharacterString> </mcc:code> </mcc:MD_Identifier> </mrs:referenceSystemIdentifier> </mrs:MD_ReferenceSystem> <bands> <mrc:name> <mcc:MD_Identifier> <mcc:code> <gco:CharacterString> red </gco:CharacterString> </mcc:code> </mcc:MD_Identifier> </mrc:name> </bands> <bands> <mrc:name> <mcc:MD_Identifier> <mcc:code> <gco:CharacterString> green </gco:CharacterString> </mcc:code> </mcc:MD_Identifier> </mrc:name> </bands> <bands> <mrc:name> <mcc:MD_Identifier> <mcc:code> <gco:CharacterString> blue </gco:CharacterString> </mcc:code> </mcc:MD_Identifier> </mrc:name> </bands> <imageSize> 6000x7600 </imageSize> <tasks> <type> AI_EOTask </type> <id> whu_rs19-task </id> <description> Structural high-resolution satellite image indexing </description> <taskType> Scene Classification </taskType> </tasks> <data> <type> AI_EOTrainingData </type> <id> airport_01 </id> <dataSources> <cit:title> <gco:CharacterString> googleEarth </gco:CharacterString> </cit:title> </dataSources> <dataURL> image/Airport/airport_01.jpg </dataURL> <labels> <type> AI_SceneLabel </type> <class> Airport </class> </labels> </data> 7.3 . Requirements Class: AI_TrainingData The AI_TrainingData requirements class defines an XML encoding for the AI_TrainingData module, which is based on the UML model specified in the TrainingDML-AI Part 1: Conceptual Model Standard. Requirements class /req/aitrainingdata Dependency XML Dependency /req/base/xmlbasetype Dependency /req/base/isometadatatype Dependency /req/ailabel Dependency /req/ailabeling Dependency /req/aidataquality Requirement 17 /req/aitrainingdata/trainingdata Requirement 18 /req/aitrainingdata/trainingtypecode Requirement 19 /req/aitrainingdata/eotrainingdata The AI_TrainingData object is encoded as an XML element with properties shown in Table 5 . Requirement 17 /req/aitrainingdata/trainingdata Each AI_TrainingData object SHALL implement the Mandatory properties shown in Table 5 . Table 5 — AI_TrainingData properties XML Property Definition Data type and values Obligation type Type of the training sample. “AI_AbstractTrainingData” Mandatory id Identification of an individual AI training sample. CharacterString [1..1] Mandatory datasetId Identification of the training dataset that the training sample belongs to. CharacterString [0..1] Optional trainingType Training type of the individual AI training sample. AI_TrainingTypeCode [0..1] Optional numberOfLabels Total number of labels in the individual AI training sample. Int [0..1] Optional dataSources Citation of inputs to prepare a training sample. CI_Citation [0..*] Optional extent Spatial extent of the individual training sample. EX_Extent [0..1] Optional dataTime Date and time at which the geospatial data were obtained. DateTime [0..*] Optional lineage Information about the provenance, source(s), and/or the production process(es) applied to the training sample. LI_Lineage [0..*] Optional labels Labels in the training data. AI_Label [0..*] Optional labeling Provenance information of how the training data is labeled. AI_Labeling [0..*] Optional quality Quality information of the training data. DataQuality [0..*] Optional Example: <type> AI_AbstractTrainingData </type> <id> airport_01 </id> <dataSources> <cit:title> <gco:CharacterString> googleEarth </gco:CharacterString> </cit:title> </dataSources> <dataURL> image/Airport/airport_01.jpg </dataURL> <extent> <gex:geographicElement> <gex:EX_GeographicBoundingBox> <gex:westBoundLongitude> <gco:Decimal> -171.76409 </gco:Decimal> </gex:westBoundLongitude> <gex:eastBoundLongitude> <gco:Decimal> -171.56578 </gco:Decimal> </gex:eastBoundLongitude> <gex:southBoundLatitude> <gco:Decimal> -14.42443 </gco:Decimal> </gex:southBoundLatitude> <gex:northBoundLatitude> <gco:Decimal> -14.32568 </gco:Decimal> </gex:northBoundLatitude> </gex:EX_GeographicBoundingBox> </gex:geographicElement> </extent> <dataTime> 2002-05-30T09:30:10Z </dataTime> <labels> <type> AI_SceneLabel </type> <class> Airport </class> </labels> The AI_TrainingTypeCode is encoded as a text string whose value is one of “training”, “validation”, “test” or “retraining”. Requirement 18 /req/aitrainingdata/trainingtypecode Each AI_TrainingTypeCode value SHALL be a text string whose value is one of “training”, “validation”, “test” or “retraining”. Examples: a) "training" b) "validation" c) "test" d) "retraining" The AI_EOTrainingData object is encoded as an XML element with properties both shown in Table 5 and Table 6 . Requirement 19 /req/aitrainingdata/eotrainingdata Each AI_EOTrainingData object SHALL implement the Mandatory properties as defined in Table 5 and Table 6 . Table 6 — AI_EOTrainingData properties XML Property Definition Data type and values Obligation type Type of the EO training data. “AI_EOTrainingData” Mandatory spatialResolution Spatial resolution of the individual EO training sample. MD_Resolution [0..1] Optional dataURL URLs of the EO data, including both relative and absolute paths, which can encompass local paths, network addresses, and more. URI [1..*] Mandatory Example: <type> AI_EOTrainingData </type> <id> airport_01 </id> <dataSources> <cit:title> <gco:CharacterString> googleEarth </gco:CharacterString> </cit:title> </dataSources> <extent> <gex:geographicElement> <gex:EX_GeographicBoundingBox> <gex:westBoundLongitude> <gco:Decimal> -171.76409 </gco:Decimal> </gex:westBoundLongitude> <gex:eastBoundLongitude> <gco:Decimal> -171.56578 </gco:Decimal> </gex:eastBoundLongitude> <gex:southBoundLatitude> <gco:Decimal> -14.42443 </gco:Decimal> </gex:southBoundLatitude> <gex:northBoundLatitude> <gco:Decimal> -14.32568 </gco:Decimal> </gex:northBoundLatitude> </gex:EX_GeographicBoundingBox> </gex:geographicElement> </extent> <spatialResolution> <mri:MD_Resolution> <gco:Distance uom= "meter" > 10 </gco:Distance> </mri:MD_Resolution> </spatialResolution> <dataTime> 2002-05-30T09:30:10Z </dataTime> <dataURL> image/Airport/airport_01.jpg </dataURL> <labels> <type> AI_SceneLabel </type> <class> Airport </class> </labels> 7.4 . Requirements Class: AI_Task The AI_Task requirements class defines an XML encoding for the AI_Task module, which is based on the UML model specified in the TrainingDML-AI Part 1: Conceptual Model Standard. Requirements class /req/aitask Dependency XML Dependency /req/base/xmlbasetype Requirement 20 /req/aitask/task Requirement 21 /req/aitask/eotask The AI_Task object is encoded as an XML element with properties as shown in Table 7 . Requirement 20 /req/aitask/task Each AI_Task object SHALL implement the Mandatory properties shown in Table 7 . Table 7 — AI_Task properties XML Property Definition Data type and values Obligation type Type of the task object. “AI_AbstractTask” Mandatory id Identification of the task. CharacterString [1..1] Mandatory datasetId Identification of the training dataset the training sample belongs to. CharacterString [0..1] Optional description Description of the AI task. CharacterString [0..1] Optional Example: <type> AI_AbstractTask </type> <id> image-indexing-task </id> <description> Structural high-resolution satellite image indexing </description> The AI_EOTask object is encoded as an XML element with properties both shown in Table 7 and Table 8 . Requirement 21 /req/aitask/eotask Each AI_EOTask object SHALL implement the Mandatory properties shown in Table 7 and Table 8 . Table 8 — AI_EOTask properties XML Property Definition Data type and values Obligation type Type of the task object. “AI_EOTask” Mandatory taskType Type of the EO task. CharacterString [1..1] Mandatory Example: <type> AI_EOTask </type> <id> image-indexing-task </id> <description> Structural high-resolution satellite image indexing </description> <taskType> Scene Classification </taskType> 7.5 . Requirements Class: AI_Label The AI_Label requirements class defines an XML encoding for the AI_Label module, which is based on the UML model specified in the TrainingDML-AI Part 1: Conceptual Model Standard. Requirements class /req/ailabel Dependency XML Dependency /req/base/xmlbasetype Dependency /req/base/geospatialtype Requirement 22 /req/ailabel/label Requirement 23 /req/ailabel/scenelabel Requirement 24 /req/ailabel/objectlabel Requirement 25 /req/ailabel/pixellabel Requirement 26 /req/ailabel/imageformatcode The AI_Label object is encoded as an XML element with properties as shown in Table 9 . Requirement 22 /req/ailabel/label Each AI_Label object SHALL implement the Mandatory properties shown in Table 9 . Table 9 — AI_Label properties XML Property Definition Data type and values Obligation type Type of the label object. “AI_AbstractLabel” Mandatory isNegative Whether the training sample related to the label is a positive or negative sample. Bool [0..1] Default: false Optional confidence Confidence score of the labeler. Float [0..1] Default: 1.0 Range: [0, 1] Optional Example: <type> AI_AbstractLabel </type> <isNegative> false </isNegative> <confidence> 1.0 </confidence> The AI_SceneLabel object is encoded as an XML element with properties as shown in Table 10 . Requirement 23 /req/ailabel/scenelabel Each AI_SceneLabel object SHALL implement the Mandatory properties shown in Table 10 . Table 10 — AI_SceneLabel properties XML Property Definition Data type and values Obligation type Type of the label object at the scene level. “AI_SceneLabel” Mandatory class Class that records the semantic of the scene of the training sample. CharacterString [1..1] Mandatory Example: <type> AI_SceneLabel </type> <class> Airport </class> The AI_ObjectLabel object is encoded as an XML element with properties shown in Table 11 . Requirement 24 /req/ailabel/objectlabel Each AI_ObjectLabel object SHALL implement the Mandatory properties shown in Table 11 . Table 11 — AI_ObjectLabel properties XML Property Definition Data type and values Obligation type Type of the label object at the object level. “AI_ObjectLabel” Mandatory object Feature that represents the position and attributes of the object. Feature [0..1] Mandatory bboxType Type of the bbox. CharacterString [0..1] Optional class Class that records the semantic of the object type. CharacterString [1..1] Mandatory dateTime Created time of the object label. DateTime [0..1] Optional Example: <type> AI_ObjectLabel </type> <object> <gml:metaDataProperty> <gml:GenericMetaData> <truncated> 0.0 </truncated> <occluded> 0 </occluded> <alpha> -0.2 </alpha> </gml:GenericMetaData> </gml:metaDataProperty> <gml:location> <gml:GeometricComplex> <gml:element> <gml:Polygon> <gml:exterior> <gml:LinearRing> <gml:posList> 51.556272 -0.2803943 51.5562758 -0.2787397 51.5556539 -0.278736 51.5556501 -0.2803906 51.556272 -0.2803943 </gml:posList> </gml:LinearRing> </gml:exterior> </gml:Polygon> </gml:element> </gml:GeometricComplex> </gml:location> </object> <bboxType> Horizontal BBox </bboxType> <class> Pedestrian </class> The AI_PixelLabel object is encoded as an XML element with properties as shown in Table 12 . Requirement 25 /req/ailabel/pixellabel Each AI_PixelLabel object SHALL implement the Mandatory properties shown in Table 12 . Table 12 — AI_PixelLabel properties XML Property Definition Data type and values Obligation type Type of the label object at the pixel level. “AI_PixelLabel” Mandatory imageURL URL of the images representing the label information. URL [1..*] Mandatory imageFormat Image data format. AI_ImageFormatCode [1..*] Mandatory Example: <type> AI_PixelLabel </type> <imageURL> /label_5classes/GF2_PMS1__L1A0000647767-MSS1_label.tif </imageURL> <imageFormat> image/tiff; application=geotiff </imageFormat> The AI_ImageFormatCode is encoded as a text string whose value is defined in Multipurpose Internet Mail Extensions (MIME) Part Two: Media Types RFC 2046 . Requirement 26 /req/ailabel/imageformatcode Each AI_ImageFormatCode value SHALL be encoded as a text string defined in Multipurpose Internet Mail Extensions (MIME) Part Two: Media Types RFC 2046 . Examples: a) "image/tiff; application=geotiff" b) "application/x-netcdf" c) "image/png" d) "image/jp2" 7.6 . Requirements Class: AI_Labeling The AI_Labeling requirements class defines an XML encoding for the AI_Labeling module, which is based on the UML model specified in the TrainingDML-AI Part 1: Conceptual Model Standard. Requirements class /req/ailabeling Dependency XML Dependency /req/base/xmlbasetype Dependency /req/base/isometadatatype Requirement 27 /req/ailabeling/labeling Requirement 28 /req/ailabeling/labeler Requirement 29 /req/ailabeling/labelingprocedure Requirement 30 /req/ailabeling/labelingmethodcode The AI_Labeling object is encoded as an XML element with properties shown in Table 13 . Requirement 27 /req/ailabeling/labeling Each AI_Labeling object SHALL implement the Mandatory properties shown in Table 13 . Table 13 — AI_Labeling properties XML Property Definition Data type and values Obligation type Type of the labeling object. “AI_Labeling” Mandatory id Identifier of the labeling. CharacterString [1..1] Mandatory scope Description of the scope of the labeling. MD_Scope [1..1] Mandatory labelingTime Date and time at which the labeling procedure was carried out. DateTime [0..1] Optional labelers Labelers of the labeling activity. AI_Labeler [0..*] Optional procedure Procedure used in the labeling activity. AI_LabelingProcedure [0..1] Optional Example: <type> AI_Labeling </type> <id> 0 </id> <scope> <mcc:level> <mcc:MD_ScopeCode codeList= "http://standards.iso.org/iso/19115/resources/Codelist/cat/codelists.xml#MD_ScopeCode" codeListValue= "dataset" /> </mcc:level> <mcc:levelDescription> <mcc:MD_ScopeDescription> <mcc:dataset> <gco:CharacterString> whu_rs19 </gco:CharacterString> </mcc:dataset> </mcc:MD_ScopeDescription> </mcc:levelDescription> </scope> <dataTime> 2015-04-20T07:04:15Z </dataTime> <labelers> ... </labelers> <procedure> ... </procedure> The AI_Labeler object is encoded as an XML element with properties as shown in Table 14 . Requirement 28 /req/ailabeling/labeler Each AI_Labeler object SHALL implement the Mandatory properties shown in Table 14 . Table 14 — AI_Labeler properties XML Property Definition Data type and values Obligation type Type of the labeler object. “AI_Labeler” Mandatory id Identifier of the labeler. CharacterString [1..1] Mandatory name Name of the labeler. CharacterString [1..1] Mandatory Example: <type> AI_Labeler </type> <id> 0 </id> <name> Tom </name> The AI_LabelingProcedure object is encoded as an XML element with properties as shown in Table 15 . Requirement 29 /req/ailabeling/labelingprocedure Each AI_LabelingProcedure object SHALL implement the Mandatory properties shown in Table 15 . Table 15 — AI_LabelingProcedure properties XML Property Definition Data type and values Obligation type Type of the labeling procedure object. “AI_LabelingProcedure” Mandatory id Identifier of the labeling procedure. CharacterString [1..1] Mandatory methods Methods used in the labeling procedure. AI_LabelingMethodCode [1..*] Mandatory tools Tools or software used in the labeling procedure. CharacterString [0..*] Optional Example: <type> AI_LabelingProcedure </type> <id> 0 </id> <methods> manual </methods> <tools> ArcGIS </tools> The AI_LabelingMethodCode is encoded as a text string whose value is one of “manual”, “semi-automatic”, “automatic” or “unknown”. Requirement 30 /req/ailabeling/labelingmethodcode Each AI_LabelingMethodCode value SHALL be a text string whose value is one of “manual”, “semi-automatic”, “automatic” or “unknown”. Examples: a) "manual" b) "semi-automatic" c) "automatic" d) "unknown" 7.7 . Requirements Class: AI_DataQuality The AI_DataQuality requirements class defines an XML encoding for the AI_DataQuality module, which is based on the UML model specified in the TrainingDML-AI Part 1: Conceptual Model Standard. Requirements class /req/aidataquality Dependency XML Dependency /req/base/xmlbasetype Dependency /req/base/isometadatatype Dependency /req/base/isoqualitytype Requirement 31 /req/aidataquality/classbalancedegree The AI_ClassBalanceDegree object is encoded as an XML element with properties as shown in Table 16 . Requirement 31 /req/aidataquality/classbalancedegree Each AI_ClassBalanceDegree object SHALL implement the Mandatory properties as shown in Table 16 . Table 16 — AI_ClassBalanceDegree properties XML Property Definition Data type and values Obligation type Type of the class balance degree object. “AI_ClassBalanceDegree” Mandatory measure Reference to measure used. MeasureReference [1..1] Mandatory evaluationMethod Evaluation information. EvaluationMethod [1..*] Mandatory result Value obtained from applying a data quality measure. QualityResult [1..*] Mandatory Example: <type> AI_ClassBalanceDegree </type> <measure> <measureDescription> Balance degree of label classes </measureDescription> </measure> <evaluationMethod> <evaluationMethodDescription> Counting the number of training samples belonging to each class and calculating the balance degree </evaluationMethodDescription> </evaluationMethod> <result> <quantitativeResult> <value> 93.5 </value> <valueUnit> % </valueUnit> </quantitativeResult> </result> 7.8 . Requirements Class: AI_TDChangeset The AI_TDChangeset requirements class defines an XML encoding for the AI_TDChangeset module, which is based on the UML model specified in the TrainingDML-AI Part 1: Conceptual Model Standard. Requirements class /req/aitdchangeset Dependency XML Dependency /req/base/xmlbasetype Dependency /req/base/isometadatatype Dependency /req/aitrainingdata Requirement 32 /req/aitdchangeset/tdchangeset The AI_TDChangeset object is encoded as an XML element with properties shown in Table 17 . Requirement 32 /req/aitdchangeset/tdchangeset Each AI_TDChangeset object SHALL implement the Mandatory properties as shown in Table 17 . Table 17 — AI_TDChangeset properties XML Property Definition Data type and values Obligation type Type of the TD changeset object. “AI_TDChangeset” Mandatory id Identifier of the changeset. CharacterString [1..1] Mandatory datasetId Identifier of the training dataset the changeset belongs to. CharacterString [0..1] Optional version Version of the training dataset that the changeset belongs to. CharacterString [0..1] Optional changeCount Total number of changed training samples. Int [1..1] Mandatory createdTime The time that the changeset was created. DateTime [0..1] Optional add Added training samples. AI_AbstractTrainingData [0..*] Optional modify Modified training samples. AI_AbstractTrainingData [0..*] Optional delete Deleted training samples. AI_AbstractTrainingData [0..*] Optional Example: <type> AI_TDChangeset </type> <id> changeset-dota_v1.5 </id> <datasetId> dota_v1.5 </datasetId> <createdTime> 2019-01-01T00:00:00 </createdTime> <changeCount> 9 </changeCount> <modify> <type> EOTrainingData </type> <id> P1228 </id> <dataSources> <cit:title> <gco:CharacterString> GF </gco:CharacterString> </cit:title> </dataSources> <dataURL> train/images/P1228.png </dataURL> <numberOfLabels> 50 </numberOfLabels> <trainingType> training </trainingType> <labels> <type> AI_ObjectLabel </type> <object> <gml:location> <gml:GeometricComplex> <gml:element> <gml:Polygon> <gml:exterior> <gml:LinearRing> <gml:posList> 2306.0 729.0 2330.0 729.0 2330.0 744.0 2306.0 744.0 2306.0 729.0 </gml:posList> </gml:LinearRing> </gml:exterior> </gml:Polygon> </gml:element> </gml:GeometricComplex> </gml:location> </object> <bboxType> Horizontal BBox </bboxType> <class> ship </class> </labels> </modify> Annex A (normative) Abstract Test Suite Annex A Abstract Test Suite A.1 . Introduction Conformance is tested using the XML Schema document which formalize the requirements described above. A.2 . Conformance Class: Base The Base conformance class tests that occurrences of the basic types are encoded according to the requirements. A.2.1 . Conformance Class: XML Base Type Conformance class A.1: XML Base Type Identifier http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/base/xmlbasetype Requirements class /req/base/xmlbasetype Conformance tests Abstract test A.1: /conf/base/xmlbasetype/xml Abstract test A.2: /conf/base/xmlbasetype/datetime Abstract test A.3: /conf/base/xmlbasetype/namedvalue Abstract test A.4: /conf/base/xmlbasetype/url Abstract test A.1 Identifier /conf/base/xmlbasetype/xml Requirement /req/base/xmlbasetype/xml Test purpose Verify that the document is well-formed XML. Test method Load the document in an XML validator. Pass if no errors reported. Fail otherwise. Abstract test A.2 Identifier /conf/base/xmlbasetype/datetime Requirement /req/base/xmlbasetype/datetime Test purpose Verify that XML instance documents claiming conformance to this specification contain valid DateTime values according to Section 3.3.7 , Section 3.3.8 or Section 3.3.9 of W3C XML Schema Definition Language (XSD) 1.1 Part 2: Datatypes. Test method Inspect the instance document to verify the above requirement. Abstract test A.3 Identifier /conf/base/xmlbasetype/namedvalue Requirement /req/base/xmlbasetype/namedvalue Test purpose Verify that XML instance documents claiming conformance to this specification validate against the XML schema specified in http://schemas.opengis.net/trainingdml-ai/part3/1.0/namedValue.xsd . Test method Validate the instance document against the namedValue.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Abstract test A.4 Identifier /conf/base/xmlbasetype/url Requirement /req/base/xmlbasetype/url Test purpose Verify that XML instance documents claiming conformance to this specification contain valid URL values according to the W3C definition of anyURI, which represents an Internationalized Resource Identifier Reference. An anyURI value can be absolute or relative and may have an optional fragment identifier. Test method Inspect the instance document to verify the above requirement. A.2.2 . Conformance Class: ISO Metadata Type Conformance class A.2: ISO Metadata Type Identifier http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/base/isometadatatype Requirements class /req/base/isometadatatype Conformance tests Abstract test A.5: /conf/base/isometadatatype/band Abstract test A.6: /conf/base/isometadatatype/extent Abstract test A.7: /conf/base/isometadatatype/citation Abstract test A.8: /conf/base/isometadatatype/scope Abstract test A.9: /conf/base/isometadatatype/referencesystem Abstract test A.10: /conf/base/isometadatatype/lineage Abstract test A.11: /conf/base/isometadatatype/resolution Abstract test A.5 Identifier /conf/base/isometadatatype/band Requirement /req/base/isometadatatype/band Test purpose Verify that instance documents using the MD_Band XML elements validate against the XML schema specified in https://schemas.isotc211.org/19115/-1/mrc/1.3.0/content.xsd . Test method Validate the instance document against the content.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Abstract test A.6 Identifier /conf/base/isometadatatype/extent Requirement /req/base/isometadatatype/extent Test purpose Verify that instance documents using the EX_Extent XML elements validate against the XML schema specified in https://schemas.isotc211.org/19115/-1/gex/1.3.0/extent.xsd . Test method Validate the instance document against the extent.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Abstract test A.7 Identifier /conf/base/isometadatatype/citation Requirement /req/base/isometadatatype/citation Test purpose Verify that instance documents using the CI_Citation XML elements validate against the XML schema specified in https://schemas.isotc211.org/19115/-1/cit/1.3.0/citation.xsd . Test method Validate the instance document against the citation.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Abstract test A.8 Identifier /conf/base/isometadatatype/scope Requirement /req/base/isometadatatype/scope Test purpose Verify that instance documents using the MD_Scope XML elements validate against the XML schema specified in https://schemas.isotc211.org/19115/-1/mcc/1.3.0/commonClasses.xsd . Test method Validate the instance document against the commonClasses.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Abstract test A.9 Identifier /conf/base/isometadatatype/referencesystem Requirement /req/base/isometadatatype/referencesystem Test purpose Verify that instance documents using the MD_ReferenceSystem XML elements validate against the XML schema specified in https://schemas.isotc211.org/19115/-1/mcc/1.3.0/referenceSystem.xsd . Test method Validate the instance document against the commonClasses.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Abstract test A.10 Identifier /conf/base/isometadatatype/lineage Requirement /req/base/isometadatatype/lineage Test purpose Verify that instance documents using the LI_Lineage XML elements validate against the XML schema specified in https://schemas.isotc211.org/19115/-1/mcc/1.3.0/lineage.xsd . Test method Validate the instance document against the commonClasses.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Abstract test A.11 Identifier /conf/base/isometadatatype/resolution Requirement /req/base/isometadatatype/resolution Test purpose Verify that instance documents using the MD_Resolution XML elements validate against the XML schema specified in https://schemas.isotc211.org/19115/-1/mcc/1.3.0/resolution.xsd . Test method Validate the instance document against the commonClasses.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. A.2.3 . Conformance Class: ISO Quality Type Conformance class A.3: ISO Quality Type Identifier http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/base/isoqualitytype Requirements class /req/base/isoqualitytype Conformance test Abstract test A.12: /conf/base/isoqualitytype/element Abstract test A.12 Identifier /conf/base/isoqualitytype/element Requirement /req/base/isoqualitytype/element Test purpose Verify that instance documents using the QualityElement XML elements validate against the XML schema specified in https://schemas.isotc211.org/schemas/19157/-/mdq/1.2.0/dataQualityElement.xsd . Test method Validate the instance document against the dataQualityElement.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. A.2.4 . Conformance Class: Geospatial Type Conformance class A.4: Geospatial Type Identifier http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/base/geospatialtype Requirements class /req/base/geospatialtype Conformance test Abstract test A.13: /conf/base/geospatialtype/feature Abstract test A.13 Identifier /conf/base/geospatialtype/feature Requirement /req/base/geospatialtype/feature Test purpose Verify that instance documents using the Feature XML elements validate against the XML schema specified in http://schemas.opengis.net/gml/3.2.1/gml.xsd . Test method Validate the instance document against the gml.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. A.3 . Conformance Class: AI_TrainingDataset The AI_TrainingDataset conformance class tests that the training dataset object is encoded according to the requirements. Conformance class A.5: AI_TrainingDataset Identifier http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/aitrainingdataset Requirements class /req/aitrainingdataset Prerequisites Conformance class A.1: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/base/xmlbasetype Conformance class A.2: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/base/isometadatatype Conformance class A.6: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/aitrainingdata Conformance class A.7: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/aitask Conformance class A.9: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/ailabeling Conformance class A.10: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/aidataquality Conformance class A.11: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/aitdchangeset Conformance tests Abstract test A.14: /conf/aitrainingdataset/trainingdataset Abstract test A.15: /conf/aitrainingdataset/metricsinliterature Abstract test A.16: /conf/aitrainingdataset/eotrainingdataset Abstract test A.14 Identifier /conf/aitrainingdataset/trainingdataset Requirement /req/aitrainingdataset/trainingdataset Test purpose Verify that instance documents using the AI_TrainingDataset XML elements listed in Table 2 validate against the XML schema specified in http://schemas.opengis.net/trainingdml-ai/part3/1.0/ai_trainingDataset.xsd . Test method Validate the instance document against the ai_trainingDataset.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Abstract test A.15 Identifier /conf/aitrainingdataset/metricsinliterature Requirement /req/aitrainingdataset/metricsinliterature Test purpose Verify that instance documents using the AI_MetricsInLiterature XML elements listed in Table 3 validate against the XML schema specified in http://schemas.opengis.net/trainingdml-ai/part3/1.0/ai_metricsInLiterature.xsd . Test method Validate the instance document against the ai_metricsInLiterature.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Abstract test A.16 Identifier /conf/aitrainingdataset/eotrainingdataset Requirement /req/aitrainingdataset/eotrainingdataset Test purpose Verify that instance documents using the AI_EOTrainingDataset XML elements listed in Table 2 and Table 4 validate against the XML schema specified in http://schemas.opengis.net/trainingdml-ai/part3/1.0/ai_eoTrainingDataset.xsd . Test method Validate the instance document against the ai_eoTrainingDataset.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. A.4 . Conformance Class: AI_TrainingData The AI_TrainingData conformance class tests that the training data objects are encoded according to the requirements. Conformance class A.6: AI_TrainingData Identifier http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/aitrainingdata Requirements class /req/aitrainingdata Prerequisites Conformance class A.1: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/base/xmlbasetype Conformance class A.2: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/base/isometadatatype Conformance class A.8: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/ailabel Conformance class A.9: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/ailabeling Conformance class A.10: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/aidataquality Conformance tests Abstract test A.17: /conf/aitrainingdata/trainingdata Abstract test A.18: /conf/aitrainingdata/trainingtypecode Abstract test A.19: /conf/aitrainingdata/eotrainingdata Abstract test A.17 Identifier /conf/aitrainingdata/trainingdata Requirement /req/aitrainingdata/trainingdata Test purpose Verify that instance documents using the AI_TrainingData XML elements listed in Table 5 validate against the XML schema specified in http://schemas.opengis.net/trainingdml-ai/part3/1.0/ai_trainingData.xsd . Test method Validate the instance document against the ai_trainingData.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Abstract test A.18 Identifier /conf/aitrainingdata/trainingtypecode Requirement /req/aitrainingdata/trainingtypecode Test purpose Verify that instance documents using the AI_TrainingTypeCode XML elements validate against the XML schema specified in http://schemas.opengis.net/trainingdml-ai/part3/1.0/ai_trainingTypeCode.xsd . Test method Validate the instance document against the ai_trainingTypeCode.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Abstract test A.19 Identifier /conf/aitrainingdata/eotrainingdata Requirement /req/aitrainingdata/eotrainingdata Test purpose Verify that instance documents using the AI_EOTrainingData XML elements listed in Table 5 and Table 6 validate against the XML schema specified in http://schemas.opengis.net/trainingdml-ai/part3/1.0/ai_eoTrainingData.xsd . Test method Validate the instance document against the ai_eoTrainingData.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. A.5 . Conformance Class: AI_Task The AI_Task conformance class tests that the task objects are encoded according to the requirements. Conformance class A.7: AI_Task Identifier http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/aitask Requirements class /req/aitask Prerequisite Conformance class A.1: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/base/xmlbasetype Conformance tests Abstract test A.20: /conf/aitask/task Abstract test A.21: /conf/aitask/eotask Abstract test A.20 Identifier /conf/aitask/task Requirement /req/aitask/task Test purpose Verify that instance documents using the AI_Task XML elements listed in Table 7 validate against the XML schema specified in http://schemas.opengis.net/trainingdml-ai/part3/1.0/ai_task.xsd . Test method Validate the instance document against the ai_task.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Abstract test A.21 Identifier /conf/aitask/eotask Requirement /req/aitask/eotask Test purpose Verify that instance documents using the AI_EOTask XML elements listed in Table 7 and Table 8 validate against the XML schema specified in http://schemas.opengis.net/trainingdml-ai/part3/1.0/ai_eoTask.xsd . Test method Validate the instance document against the ai_eoTask.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. A.6 . Conformance Class: AI_Label The AI_Label conformance class tests that the label objects are encoded according to the requirements. Conformance class A.8: AI_Label Identifier http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/ailabel Requirements class /req/ailabel Prerequisites Conformance class A.1: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/base/xmlbasetype Conformance class A.4: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/base/geospatialtype Conformance tests Abstract test A.22: /conf/ailabel/label Abstract test A.23: /conf/ailabel/scenelabel Abstract test A.24: /conf/ailabel/objectlabel Abstract test A.25: /conf/ailabel/pixellabel Abstract test A.26: /conf/ailabel/imageformatcode Abstract test A.22 Identifier /conf/ailabel/label Requirement /req/ailabel/label Test purpose Verify that instance documents using the AI_Label XML elements listed in Table 9 validate against the XML schema specified in http://schemas.opengis.net/trainingdml-ai/part3/1.0/ai_label.xsd . Test method Validate the instance document against the ai_label.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Abstract test A.23 Identifier /conf/ailabel/scenelabel Requirement /req/ailabel/scenelabel Test purpose Verify that instance documents using the AI_SceneLabel XML elements listed in Table 10 validate against the XML schema specified in http://schemas.opengis.net/trainingdml-ai/part3/1.0/ai_sceneLabel.xsd . Test method Validate the instance document against the ai_sceneLabel.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Abstract test A.24 Identifier /conf/ailabel/objectlabel Requirement /req/ailabel/objectlabel Test purpose Verify that instance documents using the AI_ObjectLabel XML elements listed in Table 11 validate against the XML schema specified in http://schemas.opengis.net/trainingdml-ai/part3/1.0/ai_objectLabel.xsd . Test method Validate the instance document against the ai_objectLabel.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Abstract test A.25 Identifier /conf/ailabel/pixellabel Requirement /req/ailabel/pixellabel Test purpose Verify that instance documents using the AI_PixelLabel XML elements listed in Table 12 validate against the XML schema specified in http://schemas.opengis.net/trainingdml-ai/part3/1.0/ai_pixelLabel.xsd . Test method Validate the instance document against the ai_pixelLabel.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Abstract test A.26 Identifier /conf/ailabel/imageformatcode Requirement /req/ailabel/imageformatcode Test purpose Verify that instance documents using the AI_ImageFormatCode XML elements conform to the requirements specified by the Multipurpose Internet Mail Extensions (MIME) Part Two: Media Types RFC 2046 . Test method Inspect the instance document to verify the above requirement. A.7 . Conformance Class: AI_Labeling The AI_Labeling conformance class tests that the labeling objects are encoded according to the requirements. Conformance class A.9: AI_Labeling Identifier http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/ailabeling Requirements class /req/ailabeling Prerequisites Conformance class A.1: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/base/xmlbasetype Conformance class A.2: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/base/isometadatatype Conformance tests Abstract test A.27: /conf/ailabeling/labeling Abstract test A.28: /conf/ailabeling/labeler Abstract test A.29: /conf/ailabeling/labelingprocedure Abstract test A.30: /conf/ailabeling/labelingmethodcode Abstract test A.27 Identifier /conf/ailabeling/labeling Requirement /req/ailabeling/labeling Test purpose Verify that instance documents using the AI_Labeling XML elements listed in Table 13 validate against the XML schema specified in http://schemas.opengis.net/trainingdml-ai/part3/1.0/ai_labeling.xsd . Test method Validate the instance document against the ai_labeling.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Abstract test A.28 Identifier /conf/ailabeling/labeler Requirement /req/ailabeling/labeler Test purpose Verify that instance documents using the AI_Labeler XML elements listed in Table 14 validate against the XML schema specified in http://schemas.opengis.net/trainingdml-ai/part3/1.0/ai_labeler.xsd . Test method Validate the instance document against the ai_labeler.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Abstract test A.29 Identifier /conf/ailabeling/labelingprocedure Requirement /req/ailabeling/labelingprocedure Test purpose Verify that instance documents using the AI_LabelingProcedure XML elements listed in Table 15 validate against the XML schema specified in http://schemas.opengis.net/trainingdml-ai/part3/1.0/ai_labelingProcedure.xsd . Test method Validate the instance document against the ai_labelingProcedure.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Abstract test A.30 Identifier /conf/ailabeling/labelingmethodcode Requirement /req/ailabeling/labelingmethodcode Test purpose Verify that instance documents using the AI_LabelingMethodCode XML elements validate against the XML schema specified in http://schemas.opengis.net/trainingdml-ai/part3/1.0/ai_labelingMethodCode.xsd . Test method Validate the instance document against the ai_labelingMethodCode.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. A.8 . Conformance Class: AI_DataQuality The AI_DataQuality conformance class tests that the data quality objects are encoded according to the requirements. Conformance class A.10: AI_DataQuality Identifier http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/aidataquality Requirements class /req/aidataquality Prerequisites Conformance class A.1: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/base/xmlbasetype Conformance class A.2: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/base/isometadatatype Conformance class A.3: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/base/isoqualitytype Conformance test Abstract test A.31: /conf/aidataquality/classbalancedegree Abstract test A.31 Identifier /conf/aidataquality/classbalancedegree Requirement /req/aidataquality/classbalancedegree Test purpose Verify that instance documents using the AI_ClassBalanceDegree XML elements listed in Table 16 validate against the XML schema specified in http://schemas.opengis.net/trainingdml-ai/part3/1.0/ai_classBalanceDegree.xsd . Test method Validate the instance document against the ai_classBalanceDegree.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. A.9 . Conformance Class: AI_TDChangeset The AI_TDChangeset conformance class tests that the TD changeset objects are encoded according to the requirements. Conformance class A.11: AI_TDChangeset Identifier http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/aitdchangeset Requirements class /req/aitdchangeset Prerequisites Conformance class A.1: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/base/xmlbasetype Conformance class A.6: http://www.opengis.net/spec/TrainingDML-AI-3/1.0/conf/aitrainingdata Conformance test Abstract test A.32: /conf/aitdchangeset/tdchangeset Abstract test A.32 Identifier /conf/aitdchangeset/tdchangeset Requirement /req/aitdchangeset/tdchangeset Test purpose Verify that instance documents using the AI_TDChangeset XML elements listed in Table 17 validate against the XML schema specified in http://schemas.opengis.net/trainingdml-ai/part3/1.0/ai_tdChangeset.xsd . Test method Validate the instance document against the ai_tdChangeset.xsd schema to verify the above requirement. The process may be using an appropriate software tool for validation or be a manual process that checks all definitions from the XML schema specification. Annex B (informative) Example Annex B Example B.1 . TrainingDataset Encoding Examples B.1.1 . WHU-RS19 Dataset The WHU-RS19 dataset is widely used in scene classification of remote sensing images. This dataset is collected from Google Earth and has 19 classes including airport, beach, bridge, commercial, desert, farmland, football field, forest, industrial, meadow, mountain, park, parking, pond, port, railway station, residential, river, and viaduct. Each class contains around 50 images, with an image size of 600×600 and a resolution of 0.5 m. An example of XML encoding of the WHU-RS19 dataset following the TrainingDML-AI UML model can be found in https://github.com/opengeospatial/TrainingDML-AI_SWG/tree/main/use-cases/examples/1.0/WHU-RS19.xml . B.1.2 . DOTA-v1.5 Dataset The DOTA-v1.5 dataset is a large-scale dataset for object detection in aerial images. The sources for content in the dataset include Google Earth, Gaofen-2, and Jilin-1 imagery provided by China Resources Satellite Data Center. The 16 classes in DOTA-v1.5 are plane, ship, storage tank, baseball diamond, tennis court, basketball court, ground track field, harbor, bridge, large vehicle, small vehicle, helicopter, roundabout, soccer ball field, swimming pool, and container crane. Compared with other aerial image object detection datasets, the dataset has the largest number of classes. The images in the dataset have various image sizes (from 800×800 to 2000×2000) and resolutions (Google Earth/0.1 m-1 m, Gaofen-2/1 m, Jilin-1/0.72 m). An example of XML encoding of the DOTA-v1.5 dataset following the TrainingDML-AI UML model can be found in https://github.com/opengeospatial/TrainingDML-AI_SWG/tree/main/use-cases/examples/1.0/DOTA-v1.5.xml . B.1.3 . KITTI 2D Object Detection Dataset The KITTI 2D object detection dataset is a novel open-access dataset and benchmark for road area and ego-lane detection. KITTI 2D consists of 7481 annotated training images of high variability from the KITTI autonomous driving platform by two PointGrey Flea2 color cameras, capturing a broad spectrum of urban street views and road scenes. The eight (8) classes in the KITTI 2D object detection dataset are car, van, truck, pedestrian, person_sitting, cyclist, tram, and misc. Compared with other street view object detection datasets, this dataset compresses diverse scenarios and captures real-world traffic situations, ranging from freeways over rural areas to inner-city scenes with many static and dynamic objects. An example of XML encoding of the KITTI 2D object detection dataset following the TrainingDML-AI UML model can be found in https://github.com/opengeospatial/TrainingDML-AI_SWG/tree/main/use-cases/examples/1.0/KITTI.xml . B.1.4 . GID Dataset The GID dataset is one of state-of-art land cover classification datasets. This dataset has a large spatial coverage covering many provinces in China with a relatively high spatial resolution (2 m). GID has two sets. One is the GID-5C. It has 150 images (image size 7200×6800) that are classified into 5 land cover classes. The other set is GID-15C. The images from GID-5C are sliced into 30,000 patches in GID-15C, which have three types of patch sizes (56×56, 112×112, 224×224) and are classified into 15 land cover classes. An example of XML encoding of the GID-5C dataset following the TrainingDML-AI UML model can be found in https://github.com/opengeospatial/TrainingDML-AI_SWG/tree/main/use-cases/examples/1.0/GID-5C.xml . B.1.5 . Toronto3D Dataset The Toronto3D dataset is a large urban outdoor point cloud dataset for segmentation collected by the Mobile Laser Scanning System. The dataset covers about 1 km of scene streets in Toronto, including four areas named L001, L002, L003, and L004, with a total of 78.3 million points. Each point in this dataset has 10 attributes representing the 3D position, RGB color, intensity, GPS time, scan angle rank, and category, respectively. This dataset has eight categories, including road, road mark, natural, building, utility line, pole, car, and fence. An example of XML encoding of the Toronto3D dataset following the TrainingDML-AI UML model can be found in https://github.com/opengeospatial/TrainingDML-AI_SWG/tree/main/use-cases/examples/1.0/Toronto_3D.xml . B.1.6 . WHU-Building Dataset The WHU-Building dataset is a change detection dataset collected from the Land Information New Zealand Data Service. The dataset is composed of images (with the resolution 0.2 m) in 2012 and 2016, covering 20.5 km2. It includes 12,796 and 16,077 buildings respectively in 2012 and 2016. An example of XML encoding of the WHU-Building dataset following the TrainingDML-AI UML model can be found in https://github.com/opengeospatial/TrainingDML-AI_SWG/tree/main/use-cases/examples/1.0/WHU-building.xml . B.1.7 . California Change Detection Dataset The California Change Detection Dataset is composed of two images and a label image. The first image is a Landsat 8 acquisition covering Sacramento County, Yuba County and Sutter County, California, on 5 January 2017. It has nine channels covering the spectrum from deep blue to short-wave infrared, plus two long-wave infrared channels. The second image was acquired on 18 February 2017 by Sentinel-1A over the same area after the occurrence of a flood. The image is recorded in polarizations VV and VH and augmented with the ratio between the two intensities as a third channel. All these channels are log-transformed. An example of XML encoding of the California change detection dataset following the TrainingDML-AI UML model can be found in https://github.com/opengeospatial/TrainingDML-AI_SWG/tree/main/use-cases/examples/1.0/UiT_HCD_California_2017.xml . B.1.8 . WHU MVS Dataset The WHU MVS dataset is a synthetic aerial dataset created for large-scale and high-resolution Earth surface reconstruction. The basic training sample of the dataset is a multi-view unit consisting of five aerial images, and their corresponding depth maps are taken as ground truth. There are a total of 5680 pairs of five-view aerial images in the dataset. All the images are simulated from a 3D surface model, which is produced by Smart3D software using Unmanned Aerial Vehicle (UAV) images and refined by manual editing. An example of XML encoding of the WHU MVS dataset following the TrainingDML-AI UML model can be found in https://github.com/opengeospatial/TrainingDML-AI_SWG/tree/main/use-cases/examples/1.0/WHU_MVS.xml . B.1.9 . iSAID Dataset The iSAID dataset is the first benchmark dataset for instance segmentation in aerial images. This large-scale and densely annotated dataset contains 655,451 object instances for 15 categories across 2,806 high-resolution images. The images of iSAID is the same as the DOTA-v1.0 dataset, which are mainly collected from the Google Earth, some are taken by satellite JL-1, the others are taken by satellite GF-2 of the China Centre for Resources Satellite Data and Application. The object categories in iSAID include: plane, ship, storage tank, baseball diamond, tennis court, basketball court, ground track field, harbor, bridge, large vehicle, small vehicle, helicopter, roundabout, soccer ball field and swimming pool. An example of XML encoding of the iSAID dataset following the TrainingDML-AI UML model can be found in https://github.com/opengeospatial/TrainingDML-AI_SWG/blob/main/use-cases/examples/1.0/iSAID.xml . B.2 . DataQuality Encoding Example B.2.1 . WHU-RS19 Data Quality An encoded data quality example of the WHU-RS19 datasets following the TrainingDML-AI UML model can be found in https://github.com/opengeospatial/TrainingDML-AI_SWG/tree/main/use-cases/examples/1.0/WHU-RS19-quality.xml . B.3 . TDChangeset Encoding Example B.3.1 . DOTA-v1.5 Changeset DOTA-v1.5 uses the same images as DOTA-v1.0, but the extremely small instances (less than 10 pixels) are also annotated. Moreover, a new category “container crane” is added. It contains 403,318 instances in total. The number of images and dataset splits are the same as DOTA-v1.0. This version was released for the DOAI Challenge 2019 on Object Detection in Aerial Images in conjunction with IEEE CVPR 2019. An encoded changeset example between the DOTA-v1.0 and DOTA-v1.5 datasets following the TrainingDML-AI UML model can be found in https://github.com/opengeospatial/TrainingDML-AI_SWG/tree/main/use-cases/examples/1.0/DOTA-v1.5-changeset.xml . B.4 . Non-EO Imagery TrainingDataset Encoding Examples B.4.1 . ERA5 Dataset The ERA5 dataset is derived from in-situ observational data (Copernicus product), and we limit its usage scenario to the autoregression problem of time series data. Therefore, its label is the data itself. Similar to unsupervised learning, the autoregression task for time series data does not require additional labeled data. For this dataset, inheritance classes for AI_AbstractLabel are not defined, although this class is required in the existing standard (please note that these test cases are for future versions of the standard). In addition, additional attributes to support the complete representation of dataset information were added. An example of XML encoding of the ERA5 dataset following the TrainingDML-AI UML model can be found in https://github.com/opengeospatial/TrainingDML-AI_SWG/blob/main/use-cases/examples/1.0/ERA5_hourly_data.xml . B.4.2 . SCIERC Dataset The SCIERC dataset is derived from textual data, and its labels are the classification of the text. This dataset is a text classification problem, with the goal of information extraction and entity recognition. For this textual dataset, the Abstract class is inherited and AI_TextTrainingDataset, AI_TextTrainingData, AI_TextTask, and AI_EntityLabel respectively are defined. In addition, additional attributes to support the complete representation of dataset information were added. An example of XML encoding of the SCIERC dataset following the TrainingDML-AI UML model can be found in https://github.com/opengeospatial/TrainingDML-AI_SWG/blob/main/use-cases/examples/1.0/SCIRec.xml . B.4.3 . nuScenes Dataset The nuScenes dataset is a public large-scale dataset for autonomous driving developed by the team at Motional (formerly nuTonomy). The full dataset includes approximately 1.4M camera images, 390k LIDAR sweeps, 1.4M RADAR sweeps and 1.4M object bounding boxes in 40k keyframes. Although the training data may come from different domains, the 3D annotation boxes captured by numerous sensors in the same keyframe are targeted at the same object and are unique. Based on this, a 3D annotation box is used to organize each 3D object using AI_ObjectLabel. Since each training data and each 3D object require many additional attributes to be fully described, many additional attributes to provide a detailed description of the training dataset, training data, labels, etc. were added. An example of XML encoding of the nuScenes dataset following the TrainingDML-AI UML model can be found in https://github.com/opengeospatial/TrainingDML-AI_SWG/blob/main/use-cases/examples/1.0/nuScenes.xml . Annex C (informative) Revision History Annex C Revision History Table — Revision history Date Release Author Paragraph modified Description 2023-12-15 0.1 Peng Yue, Ruixiang Liu, Kaixuan Wang, Yongyi Mi All Draft for internal review. 2024-02-26 0.2 Peng Yue, Ruixiang Liu, Carl Reed, Yongyi Mi Most Merge edits and comments from Carl Reed. 2024-06-09 0.3 Peng Yue, Ruixiang Liu, Yongyi Mi Chapter 2, 4, Annex A Revisions after OAB review and public comments. 2026-02-28 1.0.1 Peng Yue, Baoxin Teng Most Revisions based on the DIS comments on ISO 19178-1. Bibliography [1] How Our Compliance Program Works, n.d. Open Geospatial Consortium [online]. Available from: https://www.ogc.org/how-our-compliance-program-works/ (accessed 11.3.24).