SIST EN ISO 19178-1:2025

SLOVENSKI STANDARD
01-november-2025
Geografske informacije - Jezik za označevanje podatkov za usposabljanje umetne
inteligence - 1. del: Konceptualni model (ISO 19178-1:2025)
Geographic information - Training data markup language for artificial intelligence - Part 1:
Conceptual model (ISO 19178-1:2025)
Information géographique - Langage de balisage des données d'entraînement pour
l'intelligence artificielle - Partie 1: Modèle conceptuel (ISO 19178-1:2025)
Ta slovenski standard je istoveten z: EN ISO 19178-1:2025
ICS:
07.040 Astronomija. Geodezija. Astronomy. Geodesy.
Geografija Geography
35.060 Jeziki, ki se uporabljajo v Languages used in
informacijski tehniki in information technology
tehnologiji
35.240.70 Uporabniške rešitve IT v IT applications in science
znanosti
2003-01.Slovenski inštitut za standardizacijo. Razmnoževanje celote ali delov tega standarda ni dovoljeno.

EN ISO 19178-1
EUROPEAN STANDARD
NORME EUROPÉENNE
June 2025
EUROPÄISCHE NORM
ICS 35.240.70
English Version
Geographic information - Training data markup language
for artificial intelligence - Part 1: Conceptual model (ISO
19178-1:2025)
Information géographique - Langage de balisage des
données d'entraînement pour l'intelligence artificielle -
Partie 1: Modèle conceptuel (ISO 19178-1:2025)
This European Standard was approved by CEN on 3 June 2025.

CEN members are bound to comply with the CEN/CENELEC Internal Regulations which stipulate the conditions for giving this
European Standard the status of a national standard without any alteration. Up-to-date lists and bibliographical references
concerning such national standards may be obtained on application to the CEN-CENELEC Management Centre or to any CEN
member.
This European Standard exists in three official versions (English, French, German). A version in any other language made by
translation under the responsibility of a CEN member into its own language and notified to the CEN-CENELEC Management
Centre has the same status as the official versions.

CEN members are the national standards bodies of Austria, Belgium, Bulgaria, Croatia, Cyprus, Czech Republic, Denmark, Estonia,
Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Norway,
Poland, Portugal, Republic of North Macedonia, Romania, Serbia, Slovakia, Slovenia, Spain, Sweden, Switzerland, Türkiye and
United Kingdom.
EUROPEAN COMMITTEE FOR STANDARDIZATION
COMITÉ EUROPÉEN DE NORMALISATION

EUROPÄISCHES KOMITEE FÜR NORMUNG

CEN-CENELEC Management Centre: Rue de la Science 23, B-1040 Brussels
© 2025 CEN All rights of exploitation in any form and by any means reserved Ref. No. EN ISO 19178-1:2025 E
worldwide for CEN national Members.

Contents Page
European foreword . 3

European foreword
This document (EN ISO 19178-1:2025) has been prepared by Technical Committee ISO/TC 211
"Geographic information/Geomatics" in collaboration with Technical Committee CEN/TC 287
“Geographic Information” the secretariat of which is held by BSI.
This European Standard shall be given the status of a national standard, either by publication of an
identical text or by endorsement, at the latest by December 2025, and conflicting national standards
shall be withdrawn at the latest by December 2025.
Attention is drawn to the possibility that some of the elements of this document may be the subject of
patent rights. CEN shall not be held responsible for identifying any or all such patent rights.
Any feedback and questions on this document should be directed to the users’ national standards
body/national committee. A complete listing of these bodies can be found on the CEN website.
According to the CEN-CENELEC Internal Regulations, the national standards organizations of the
following countries are bound to implement this European Standard: Austria, Belgium, Bulgaria,
Croatia, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland,
Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Norway, Poland, Portugal, Republic of
North Macedonia, Romania, Serbia, Slovakia, Slovenia, Spain, Sweden, Switzerland, Türkiye and the
United Kingdom.
Endorsement notice
The text of ISO 19178-1:2025 has been approved by CEN as EN ISO 19178-1:2025 without any
modification.
International
Standard
ISO 19178-1
First edition
Geographic information — Training
2025-05
data markup language for artificial
intelligence —
Part 1:
Conceptual model
Information géographique — Langage de balisage des données
d'entraînement pour l'intelligence artificielle —
Partie 1: Modèle conceptuel
Reference number
ISO 19178-1:2025(en) © ISO 2025

ISO 19178-1:2025(en)
© ISO 2025
All rights reserved. Unless otherwise specified, or required in the context of its implementation, no part of this publication may
be reproduced or utilized otherwise in any form or by any means, electronic or mechanical, including photocopying, or posting on
the internet or an intranet, without prior written permission. Permission can be requested from either ISO at the address below
or ISO’s member body in the country of the requester.
ISO copyright office
CP 401 • Ch. de Blandonnet 8
CH-1214 Vernier, Geneva
Phone: +41 22 749 01 11
Email: copyright@iso.org
Website: www.iso.org
Published in Switzerland
ii
ISO 19178-1:2025(en)
Contents Page
Foreword .v
Introduction .vi
1 Scope . 1
2 Normative references . 1
3 Terms, definitions and abbreviated terms . 1
3.1 Terms and definitions .1
3.2 Abbreviated terms .3
4 Conventions . 4
4.1 General .4
4.2 Identifiers .4
4.3 UML notation.4
5 Conformance . 6
6 Overview . 6
6.1 General .6
6.2 AI tasks for EO.6
6.3 Modularization .7
6.4 General modelling principles .8
6.4.1 Element modelling .8
6.4.2 Class hierarchy and inheritance of properties and relations .8
6.4.3 Definition of the semantics for all classes, properties and relations . .9
6.4.4 Data integrity, authenticity and non-repudiation .9
6.5 Extending TrainingDML-AI .9
7 TrainingDML-AI UML model . 9
7.1 General .9
7.2 ISO dependencies .9
7.3 Overview of the UML model .10
7.4 AI_TrainingDataset . 12
7.4.1 General . 12
7.4.2 Provisions . 13
7.4.3 Class definitions .14
7.5 AI_TrainingData . 15
7.5.1 General . 15
7.5.2 Provisions .16
7.5.3 Class definitions .16
7.6 AI_Task .17
7.6.1 General .17
7.6.2 Provisions .18
7.6.3 Class definitions .18
7.7 AI_Label .18
7.7.1 General .18
7.7.2 Provisions .19
7.7.3 Class definitions . 20
7.8 AI_Labeling . 20
7.8.1 General . 20
7.8.2 Provisions .21
7.8.3 Class definitions . 22
7.9 AI_TDChangeset . 22
7.9.1 General . 22
7.9.2 Provisions . 23
7.9.3 Class definitions .24
7.10 AI_DataQuality .24
7.10.1 General .24

iii
ISO 19178-1:2025(en)
7.10.2 Provisions . 25
7.10.3 Class definitions . 26
8 TrainingDML-AI Data Dictionary .26
8.1 General . 26
8.2 ISO Classes . 26
8.2.1 Feature (from ISO 19101-1) . 26
8.2.2 MD_Band (from ISO 19115-1) . 26
8.2.3 MD_Scope (from ISO 19115-1) . 26
8.2.4 MD_ReferenceSystem (from ISO 19115-1) . 26
8.2.5 LI_Lineage (from ISO 19115-1) .27
8.2.6 EX_Extent (from ISO 19115-1) .27
8.2.7 CI_Citation (from ISO 19115-1) .27
8.2.8 MD_Resolution (from ISO 19115-1) .27
8.2.9 DataQuality (from ISO 19157-1).27
8.2.10 QualityElement (from ISO 19157-1) . 28
8.3 AI_TrainingDataset . 28
8.3.1 Metadata . 28
8.3.2 Classes . 28
8.4 AI_TrainingData . 30
8.4.1 Metadata . 30
8.4.2 Classes . 30
8.5 AI_Task .31
8.5.1 Metadata .31
8.5.2 Classes .32
8.6 AI_Label .32
8.6.1 Metadata .32
8.6.2 Classes .32
8.7 AI_Labeling . 34
8.7.1 Metadata . 34
8.7.2 Classes . 34
8.8 AI_TDChangeset . 35
8.8.1 Metadata . 35
8.8.2 Classes . 35
8.9 AI_DataQuality . 36
8.9.1 Metadata . 36
8.9.2 Classes . 36
Annex A (normative) Abstract test suite .37
Annex B (informative) Examples .44
Bibliography . 47

iv
ISO 19178-1:2025(en)
Foreword
ISO (the International Organization for Standardization) is a worldwide federation of national standards
bodies (ISO member bodies). The work of preparing International Standards is normally carried out through
ISO technical committees. Each member body interested in a subject for which a technical committee
has been established has the right to be represented on that committee. International organizations,
governmental and non-governmental, in liaison with ISO, also take part in the work. ISO collaborates closely
with the International Electrotechnical Commission (IEC) on all matters of electrotechnical standardization.
The procedures used to develop this document and those intended for its further maintenance are described
in the ISO/IEC Directives, Part 1. In particular, the different approval criteria needed for the different types
of ISO document should be noted. This document was drafted in accordance with the editorial rules of the
ISO/IEC Directives, Part 2 (see www.iso.org/directives).
ISO draws attention to the possibility that the implementation of this document may involve the use of (a)
patent(s). ISO takes no position concerning the evidence, validity or applicability of any claimed patent
rights in respect thereof. As of the date of publication of this document, ISO had not received notice of (a)
patent(s) which may be required to implement this document. However, implementers are cautioned that
this may not represent the latest information, which may be obtained from the patent database available at
www.iso.org/patents. ISO shall not be held responsible for identifying any or all such patent rights.
Any trade name used in this document is information given for the convenience of users and does not
constitute an endorsement.
For an explanation of the voluntary nature of standards, the meaning of ISO specific terms and expressions
related to conformity assessment, as well as information about ISO’s adherence to the World Trade
Organization (WTO) principles in the Technical Barriers to Trade (TBT), see www.iso.org/iso/foreword.html.
This document was prepared by Technical Committee ISO/TC 211, Geographic information/Geomatics, in
collaboration with the European Committee for Standardization (CEN) Technical Committee CEN/TC 287,
Geographic Information, in accordance with the Agreement on technical cooperation between ISO and CEN
(Vienna Agreement) and in collaboration with the Open Geospatial Consortium (OGC).
A list of all parts in the ISO 19178 series can be found on the ISO website.
Any feedback or questions on this document should be directed to the user’s national standards body. A
complete listing of these bodies can be found at www.iso.org/members.html.

v
ISO 19178-1:2025(en)
Introduction
This document aims to develop the UML model and encodings for geospatial machine learning training data.
Training data play a fundamental role in Earth Observation (EO) Artificial Intelligence Machine Learning
(AI/ML), especially Deep Learning (DL). Training data are used to train, validate and test AI/ML models.
This document defines a UML model and encodings consistent with the OGC Standards baseline to exchange
and retrieve training data in the Web environment.
This document provides detailed metadata for formalizing the information model of training data. This
includes, but is not limited to the following aspects:
— how the training data are prepared, such as provenance or quality;
— how to specify different metadata used for different ML tasks, such as scene/object/pixel levels;
— how to differentiate the high-level training data information model and extended information models
specific to various ML applications;
— how to introduce external classification schemes and flexible means for representing labelling.

vi
International Standard ISO 19178-1:2025(en)
Geographic information — Training data markup language
for artificial intelligence —
Part 1:
Conceptual model
1 Scope
Within the context of training data for Earth Observation (EO) Artificial Intelligence Machine Learning (AI/
ML), this document specifies a conceptual model that:
— establishes a UML model with a target of maximizing the interoperability and usability of EO imagery
training data;
— specifies different AI/ML tasks and labels in EO in terms of supervised learning, including scene level,
object level and pixel level tasks;
— describes the permanent identifier, version, licence, training data size, measurement or imagery used for
annotation;
— specifies a description of quality (e.g. training data errors, training data representativeness, quality
measures) and provenance (e.g. agents who perform the labelling, labelling procedure).
2 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.
ISO 19101-1, Geographic information — Reference model — Part 1: Fundamentals
ISO 19103, Geographic information — Conceptual schema language
ISO 19115-1, Geographic information — Metadata — Part 1: Fundamentals
ISO 19156, Geographic information — Observations, measurements and samples
ISO 19157-1, Geographic information — Data quality — Part 1: General requirements
3 Terms, definitions and abbreviated terms
3.1 Terms and definitions
For the purposes of this document, the following terms and definitions apply.
ISO and IEC maintain terminology databases for use in standardization at the following addresses:
— ISO Online browsing platform: available at https:// www .iso .org/ obp
— IEC Electropedia: available at https:// www .electropedia .org/

ISO 19178-1:2025(en)
3.1.1
3D model reconstruction
task in which 3D objects and scenes are built from multi-view images
3.1.2
artificial intelligence
AI
branch of computer science devoted to developing data processing systems that perform functions normally
associated with human intelligence, such as reasoning, learning and self-improvement
[SOURCE: ISO/IEC 2382:2015, 2121393, modified — Notes 1 and 2 to entry have been removed.]
3.1.3
change detection
recognition of changes between images acquired at different times
3.1.4
class
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]
3.1.5
dataset
identifiable collection of data
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.
[SOURCE: ISO 19115-1:2014, 4.3]
3.1.6
deep learning
DL
approach to creating rich hierarchical representations through the training of
neural networks with one or more hidden layers
Note 1 to entry: Deep learning uses multi-layered networks of simple computing units (or “neurons”). In these neural
networks each unit combines a set of input values to produce an output value, which in turn is passed on to other
neurons downstream.
[SOURCE: ISO/IEC TR 29119-11:2020, 3.1.26]
3.1.7
generative model
method of large model training, which improves model performance through
unsupervised pre-training
Note 1 to entry: In the fine-tuning phase, labelled data play a critical role in optimizing the model for specific vertical
domains or tasks. By incorporating labelled 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 labelled data can significantly improve the performance of large models in specialized domains
or tasks.
3.1.8
label
known or expected results annotated as values of a dependent variable in training samples
Note 1 to entry: A training sample label is different from those on a geographical map, which are known as map labels
or annotations.
ISO 19178-1:2025(en)
3.1.9
machine learning
ML
process of optimizing model parameters through computational techniques, such
that the model’s behaviour reflects the data or experience
Note 1 to entry: 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: ISO/IEC 22989:2022, 3.3.5, modified — Note 1 has been added.]
3.1.10
object detection
recognition of objects from images
Note 1 to entry: The objects are often localized using bounding boxes.
3.1.11
provenance
organization or individual that created, accumulated, maintained and used records
Note 1 to entry: In this document provenance is a record of how training data were prepared.
[SOURCE: ISO 19115-1:2014, 4.16, modified —Note 1 to entry has been added.]
3.1.12
quality
degree to which a set of inherent characteristics of an object fulfils requirements
Note 1 to entry: Quality of training data (such as data imbalance and mislabelling) can impact the performance of
artificial intelligence/machine learning (AI/ML) models.
[SOURCE: ISO 9000:2015, 3.6.2, modified — Notes 1 and 2 to entry have been removed, and a new Note 1 to
entry has been added.]
3.1.13
scene classification
task of identifying scene categories of images, on the basis of a training set of images
whose scene categories are known
3.1.14
semantic segmentation
task of assigning class labels to pixels of images or points of point clouds
3.1.15
training dataset
collection of samples, often labelled with known or expected values for supervised
learning
Note 1 to entry: A training dataset can be divided into training, validation and test sets. "Training samples" referred
to in this document are different from "samples" referred to in ISO 19156. They are often collected in purposive ways
that deviate from purely probability sampling, with known or expected results labelled as values of a dependent
variable for generating a trained predictive model.
3.2 Abbreviated terms
In this document, the following abbreviated terms and acronyms are used or introduced:

ISO 19178-1:2025(en)
ATS abstract test suite
DML Data Markup Language
EO earth observation
ISO International Organization for Standardization
JSON JavaScript Object Notation
LC land cover
LU land use
OGC Open Geospatial Consortium
RS remote sensing
SAR synthetic aperture radar
TD training data
UML Unified Modelling Language
URL Uniform Resource Locator
URI Uniform Resource Identifier
XML Extensible Markup Language
4 Conventions
4.1 General
This clause provides details and examples for any conventions used in the document. Examples of conventions
are symbols, abbreviations, use of XML schema, or special notes regarding how to read the document.
4.2 Identifiers
The requirements in this specification are denoted by the URI:
http://www.opengis.net/spec/TrainingDML-AI-1/1.0

All requirements and conformance tests that appear in this document are denoted by partial URIs which are
relative to this base.
4.3 UML notation
The conceptual model is presented in this document through diagrams using the Unified Modelling Language
(UML) static structure diagram. The UML notations used in this document are described in the diagram in
Figure 1.
ISO 19178-1:2025(en)
NOTE For further information on the UML notation, see ISO 19103.
Figure 1 — UML notation
All associations between model elements in the TrainingDML-AI conceptual model are uni-directional. Thus,
associations in the model are navigable in only one direction. The direction of navigation is depicted by an
arrowhead. In general, the context an element takes within the association is indicated by its role. The role is
displayed near the target of the association. But, if the graphical representation is ambiguous, the position of
the role has to be drawn to the element to which the association points.
The following stereotypes are used in this model.
— «DataType» defines a set of properties that lack identity. A data type is a classifier with no operations,
whose primary purpose is to hold information.
— «CodeList» enumerates the valid attribute values. In contrast to Enumeration, the list of values is open
and, thus, not given inline in the TrainingDML-AI UML Model. The allowed values can be provided within
an external code list.
ISO 19178-1:2025(en)
5 Conformance
This document defines a conceptual model that is independent of any encoding or formatting technologies.
The standardization target for this document is:
— TrainingDML-AI conceptual model
Conformance with this document shall be checked using all the relevant tests specified in Annex A of this
document. 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
[9]
Testing web site.
All requirements-classes and conformance-classes described in this document are owned by the standard
identified.
6 Overview
6.1 General
This document defines how to represent and exchange ML training data. The conceptual model includes the
most relevant training data entities from datasets, to instances (i.e. individual training samples), to labels.
The conceptual schema specifies how and into which parts the training data should be decomposed and
classified.
This document strategically addresses geospatial requirements by providing a modular and extensible
framework tailored to EO applications. The content and format of training datasets differ depending on the
EO ML scenarios they were collected for (e.g. scene/object/pixel levels). This document defines a UML model
and encodings consistent with the OGC/ISO baseline standards to exchange and retrieve geospatial training
data. Existing geospatial standards (e.g. ISO 19101-1, ISO 19115-1, ISO 19157-1) can be reused when defining
geospatial requirements on source RS images, label geometry, metadata and quality. While some general
geospatial information such as the spatial extent and reference system information are defined for training
data at the high level, other EO-specific information, such as the size of each sample image, spatial resolution,
and bands, can be extended in a subclass at the low level. With a hierarchical and extensible structure,
the training data model accommodates diverse geospatial data characteristics, ensuring flexibility and
interoperability.
The training data model defined in this document facilitates interoperability by enabling heterogeneous
training datasets to conform to a unified representation and exchange form. It ensures that training
data from different vendors can be consistently shared and interpreted, improving the accessibility and
promoting the integration of geospatial AI/ML resources.
The TrainingDML-AI conceptual model (Clause 7) is formally specified using UML class diagrams,
complemented by a data dictionary (Clause 8) providing the definitions and explanations of the object
classes and attributes. This conceptual model provides the basis for specifying encoding implemented in
languages such as JSON, or XML. Annex B provides a series of encoding examples, including representations
for TrainingDataset, DataQuality, and TDChangeset encoding.
6.2 AI tasks for EO
In recent years AI/ML has been increasingly used in the EO domain. The new AI/ML algorithms frequently
require large training datasets as benchmarks. AI/ML TD have been used in many EO applications to
calibrate the performance of AI/ML models. Many efforts have been made to produce training datasets
to make accurate predictions. As a result, a number of training datasets are publicly available, with new
datasets being constantly released. In the EO domain, examples of AI/ML training datasets have been
developed in various tasks including the following typical scenarios.
— Scene classification: These algorithms determine image categories from numerous pictures (e.g.
agricultural, forest and beach scenes). The training samples are a series of labelled pictures. The data

ISO 19178-1:2025(en)
can be either from satellite, drones or aircrafts. The metadata of the datasets often includes the number
of training samples, the number of classes and the image size.
— Object detection: These algorithms detect and localize different objects (e.g. airplanes, cars and buildings)
in a single image. The image can be optical or non-optical, such as synthetic aperture radar (SAR). Recent
work also suggests an increasing focus on object detection from street view imagery. Objects can be
labelled using either polygons or bounding boxes. The bounding boxes can be either oriented vertically or
horizontally. The geometry of a bounding box can be expressed using top-left/bottom-right coordinates,
coordinates of four corners, or centre coordinates along with the length and width of the box.
— Semantic segmentation: In terms of land cover (LC) and land use (LU) classification, this process assigns
a LC/LU class label to a pixel (or groups of pixels) of RS imagery. In the context of semantic segmentation
of 3D point clouds, it classifies points of a 3D point cloud into categories. TDs are usually composed of RS
images/point clouds, and the corresponding labelled value of each pixel/point recording its class.
— Change detection: These algorithms identify the difference between images acquired over the same
geographical area but taken at different times. The TD comprise a set of pre-change and post-change RS
images, with the corresponding reference map labelled for changed and unchanged pixels. The image
can be optical or SAR images.
— 3D model reconstruction: These algorithms infer the 3D geometry and structure of objects and scenes,
mainly realized from the dense matching of multi-view images. The TD are usually composed of two-
view or multi-view i
...