AS/NZS ISO 19178.1:2025 identically adopts ISO 19178 1:2025, which specifies a conceptual model within the context of training data for Earth Observation (EO) Artificial Intelligence Machine Learning (AI/ML) that establishes a UML model to maximize the interoperability and usability of EO imagery training data, and specifies different AI/ML tasks and labels in EO in terms of supervised learning, describes the permanent identifier, version, licence, training data size, measurement or imagery used for annotation, and specifies a description of quality and provenance
Table of contents
Header
About this publication
Preface
Foreword
Introduction
1 Scope
2 Normative references
3 Terms, definitions and abbreviated terms
3.1 Terms and definitions
3.2 Abbreviated terms
4 Conventions
4.1 General
4.2 Identifiers
4.3 UML notation
5 Conformance
6 Overview
6.1 General
6.2 AI tasks for EO
6.3 Modularization
6.4 General modelling principles
6.4.1 Element modelling
6.4.2 Class hierarchy and inheritance of properties and relations
6.4.3 Definition of the semantics for all classes, properties and relations
6.4.4 Data integrity, authenticity and non-repudiation
6.5 Extending TrainingDML-AI
7 TrainingDML-AI UML model
7.1 General
7.2 ISO dependencies
7.3 Overview of the UML model
7.4 AI_TrainingDataset
7.4.1 General
7.4.2 Provisions
7.4.3 Class definitions
7.5 AI_TrainingData
7.5.1 General
7.5.2 Provisions
7.5.3 Class definitions
7.6 AI_Task
7.6.1 General
7.6.2 Provisions
7.6.3 Class definitions
7.7 AI_Label
7.7.1 General
7.7.2 Provisions
7.7.3 Class definitions
7.8 AI_Labeling
7.8.1 General
7.8.2 Provisions
7.8.3 Class definitions
7.9 AI_TDChangeset
7.9.1 General
7.9.2 Provisions
7.9.3 Class definitions
7.10 AI_DataQuality
7.10.1 General
7.10.2 Provisions
7.10.3 Class definitions
8 TrainingDML-AI Data Dictionary
8.1 General
8.2 ISO Classes
8.2.1 Feature (from ISO 19101-1)
8.2.2 MD_Band (from ISO 19115-1)
8.2.3 MD_Scope (from ISO 19115-1)
8.2.4 MD_ReferenceSystem (from ISO 19115-1)
8.2.5 LI_Lineage (from ISO 19115-1)
8.2.6 EX_Extent (from ISO 19115-1)
8.2.7 CI_Citation (from ISO 19115-1)
8.2.8 MD_Resolution (from ISO 19115-1)
8.2.9 DataQuality (from ISO 19157-1)
8.2.10 QualityElement (from ISO 19157-1)
8.3 AI_TrainingDataset
8.3.1 Metadata
8.3.2 Classes
8.3.2.1 AI_AbstractTrainingDataset
8.3.2.2 AI_EOTrainingDataset
8.3.2.3 AI_MetricsInLiterature
8.4 AI_TrainingData
8.4.1 Metadata
8.4.2 Classes
8.4.2.1 AI_AbstractTrainingData
8.4.2.2 AI_EOTrainingData
8.4.2.3 AI_TrainingTypeCode
8.5 AI_Task
8.5.1 Metadata
8.5.2 Classes
8.5.2.1 AI_AbstractTask
8.5.2.2 AI_EOTask
8.6 AI_Label
8.6.1 Metadata
8.6.2 Classes
8.6.2.1 AI_AbstractLabel
8.6.2.2 AI_SceneLabel
8.6.2.3 AI_ObjectLabel
8.6.2.4 AI_PixelLabel
8.7 AI_Labeling
8.7.1 Metadata
8.7.2 Classes
8.7.2.1 AI_Labeling
8.7.2.2 AI_Labeler
8.7.2.3 AI_LabelingProcedure
8.8 AI_TDChangeset
8.8.1 Metadata
8.8.2 Classes
8.9 AI_DataQuality
8.9.1 Metadata
8.9.2 Classes
Annex A
A.1 Introduction
A.2 Conformance class AI_TrainingDataset
A.3 Conformance class AI_TrainingData
A.4 Conformance class AI_Task
A.5 Conformance class AI_Label
A.6 Conformance class AI_Labeling
A.7 Conformance class AI_TDChangeset
A.8 Conformance class AI_DataQuality
Annex B
B.1 TrainingDataset encoding examples
B.1.1 WHU-RS19 dataset
B.1.2 DOTA-v1.5 dataset
B.1.3 KITTI 2D object detection dataset
B.1.4 GID dataset
B.1.5 Toronto3D dataset
B.1.6 WHU-Building dataset
B.1.7 California change detection dataset
B.1.8 WHU MVS dataset
B.1.9 AiRound-aerial dataset
B.1.10 COWC dataset
B.1.11 MBD dataset
B.2 DataQuality encoding example — WHU-RS19 Data Quality
B.3 TDChangeset encoding example — DOTA-v1.5 Changeset
