I.  Abstract

The OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 2: JSON Encoding Standard defines requirements for encoding AI training datasets as JavaScript Object Notation (JSON). JSON is widely used for encoding data in Web-based applications. It consists of sets of objects described by name/value pairs. TrainingDML-AI Part 2 is based on the OGC Training Data Markup Language for Artificial Intelligence (TrainingDML-AI) Part 1: Conceptual Model Standard.

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.

1.  Scope

This OGC TrainingDML — AI Part 2: JSON Encoding Standard defines a JSON encoding for the exchange of training datasets. The TrainingDML — AI Part 2 Standard provides a JSON-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 document defines a JSON encoding for AI training datasets. The standardization target for this Standard is:

Conformance with this Standard 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 Testing website.

All requirements-classes and conformance-classes described in this document are owned by the standard identified.

This document 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.

IETF: RFC 7159, The JavaScript Object Notation (JSON) Data Interchange Format, 2014

IETF: RFC 7946, The GeoJSON Format, 2016

IETF: RFC 3986, Uniform Resource Identifiers (URI): Generic Syntax, 2005

IETF: RFC 3339, Date and Time on the Internet: Timestamps, 2002

IETF: RFC 2046, Multipurpose Internet Mail Extensions (MIME) Part Two: Media Types, 1996

ISO/IEC 21778:2017 Information technology — The JSON data interchange syntax

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 19157-1:2023 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.

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

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

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

identifiable collection of data

[SOURCE: ISO 19115‑1:2014, 4.3]

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

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

<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]

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

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/)

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

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)

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

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

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

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

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

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

is a lightweight, text-based, language-independent syntax for defining data interchange formats. It was derived from the ECMAScript programming language but is programming language independent. JSON defines a small set of structuring rules for the portable representation of structured data.

SOURCE: ISO/IEC 21778:2017

is a vocabulary that allows you to annotate and validate JSON documents.

SOURCE: https://json-schema.org/

refers to the entity or individual responsible for creating and releasing the JSON-based serialization syntax for geospatial training datasets, as defined in the TrainingDML-AI Part 2: JSON Encoding Standard.

5.  Conventions

This section provides details and examples for any conventions used in the document.

6.  Overview

The TrainingDML-AI Part 2: JSON Encoding Standard defines a JSON-based serialization syntax for geospatial training datasets. While other serialization formats are possible, such alternatives are not discussed in this Standard.

JSON does not have a formal class model. JSON objects are just sets of properties. However, the JSON encoding described in this Standard features a “type” property on each JSON object.

A training dataset document conforming to this Standard is a JSON document whose root value is an AI_TrainingDataset object.

7.  Requirements for TrainingDML-AI JSON Encoding

 a)	"2002-05-30T09:30:10Z"
 b)	"2002-09-24"

a) {
     "key": "forest",
     "value": "RGB(0,255,255)"
   }
b) {
     "key": "precision",
     "value": 0.8
   }

 a)	"http://www.opengeospatial.org"
 b)	"/file.txt"

a) {
     "name": [
       {
         "code": "red"
       }
     ]
   }
b) {
     "name": [
       {
         "code": "B4"
       }
     ]
   }
c) {
     "boundMax": 690,
     "boundMin": 630,
     "boundUnits": "nm"
   }

a) [
     120.0,
     30.0,
     130.0,
     40.0
   ]
b) [
     120.0,
     30.0,
     10.0,
     130.0,
     40.0,
     20.0
   ]
c) {
     "geographicElement": [
       {
         "westBoundLongitude": -171.76409,
         "eastBoundLongitude": -157.86768,
         "southBoundLatitude": -14.42443,
         "northBoundLatitude": 21.31573
       }
     ],
     "crs": "EPSG:4326"
   }
d) {
     "geographicElement": [
       {
         "polygon": [
           {
             "exterior": {
               "LinearRing": {
                 "posList": [
                   51.556272,
                   -0.2803943,
                   51.5562758,
                   -0.2787397,
                   51.5556539,
                   -0.278736,
                   51.5556501,
                   -0.2803906,
                   51.556272,
                   -0.2803943
                 ]
               }
             }
           }
         ]
       }
     ],
     "crs": "EPSG:4326"
   }

{
  "title": "Open Geospatial Consortium",
  "alternateTitle": [
    "OGC"
  ],
  "identifier": [
    {
      "code": "https://portal.ogc.org/files/?artifact_id=104605&version=1"
    }
  ]
}

a) {
     "level": "dataset",
     "levelDescription": [
       {
         "dataset": "whu_rs19"
       }
     ]
   }
b) {
     "level": "attribute",
     "levelDescription": [
       {
         "attributes": "completeness"
       }
     ]
   }
c) {
     "level": "coverage",
     "extent": [
       {
         "geographicElement": [
           {
             "westBoundLongitude": -171.76409,
             "eastBoundLongitude": -157.86768,
             "southBoundLatitude": -14.42443,
             "northBoundLatitude": 21.31573
           }
         ]
       }
     ]
   }
d) {
     "level": "feature",
     "extent": [
       {
         "geographicElement": [
           {
             "polygon": [
               {
                 "exterior": {
                   "LinearRing": {
                     "posList": [
                       51.556272,
                       -0.2803943,
                       51.5562758,
                       -0.2787397,
                       51.5556539,
                       -0.278736,
                       51.5556501,
                       -0.2803906,
                       51.556272,
                       -0.2803943
                     ]
                   }
                 }
               }
             ]
           }
         ]
       }
     ]
   }

a) {
     "referenceSystem": {
       "referenceSystemIdentifier": "EPSG:3857"
     }
   }
b) {
     "referenceSystem": {
       "referenceSystemIdentifier": "EPSG:4326"
     }
   }

{
  "LI_Lineage": {
    "statement": "The dataset was generated from high-resolution satellite imagery through a standardized preprocessing and manual labeling workflow for AI model training.",
    "source": [
      {
        "LI_Source": {
          "description": "High-resolution multispectral satellite imagery acquired in 2023 covering urban areas.",
          "sourceCitation": {
            "CI_Citation": {
              "title": "Urban Satellite Imagery Dataset 2023"
            }
          }
        }
      }
    ],
    "processStep": [
      {
        "LI_ProcessStep": {
          "description": "Radiometric calibration and geometric correction were applied to the raw satellite images.",
          "rationale": "To ensure spatial alignment and consistent reflectance values for subsequent labeling.",
          "stepDateTime": {
            "TimeInstant": "2023-06-15T10:00:00Z"
          }
        }
      },
      {
        "LI_ProcessStep": {
          "description": "Trained annotators manually labeled buildings and roads using a GIS-based annotation platform.",
          "rationale": "To generate high-quality ground truth data for supervised deep learning models.",
          "stepDateTime": {
            "TimePeriod": {
              "begin": "2023-06-20",
              "end": "2023-07-10"
            }
          }
        }
      }
    ]
  }
}

a) {
     "distance": {
       "value": 0.5,
       "uom": "m"
     }
   }
b) {
     "equivalentScale": {
       "MD_RepresentativeFraction": {
         "denominator": 10000
       }
     }
   }
c) {
     "vertical": {
       "value": 1,
       "uom": "m"
     }
   }
d) {
     "angularDistance": {
       "value": 0.01,
       "uom": "degree"
     }
   }
e) {
     "levelOfDetail": "High resolution imagery suitable for urban analysis"
   }

{
  "type": "FormatConsistency",
  "measure": {
    "measureDescription": "Percentage of training samples with inconsistent image format"
  },
  "evaluationMethod": [
    {
      "evaluationMethodDescription": "Full test method to calculate the percentage of training samples with an inconsistent format"
    }
  ],
  "result": [
    {
      "quantitativeResult": {
        "value": [
          0
        ],
        "valueUnit": "%"
      }
    }
  ]
}

a) {
     "type": "Feature",
     "geometry": {
       "type": "Point",
       "coordinates": [
         120.0,
         30.0
       ]
     },
     "properties": {
       "class": "station"
     }
   }
b) {
     "type": "Feature",
     "geometry": {
       "type": "LineString",
       "coordinates": [
         [
           120.0,
           30.0
         ],
         [
           130.0,
           40.0
         ]
       ]
     },
     "properties": {
       "class": "road"
     }
   }
c) {
     "type": "Feature",
     "geometry": {
       "type": "Polygon",
       "coordinates": [
         [
           [
             120.0,
             30.0
           ],
           [
             130.0,
             30.0
           ],
           [
             125.0,
             40.0
           ],
           [
             120.0,
             30.0
           ]
         ]
       ]
     },
     "properties": {
       "class": "building"
     }
   }

{
  "type": "AI_AbstractTrainingDataset",
  "id": "whu_rs19",
  "name": "WHU-RS19",
  "description": "Wuhan University-Remote Sensing 19 Categories (WHU-RS19) has 19 classes of remote sensing images scenes obtained from Google Earth",
  "license": "CC BY-SA 4.0",
  "amountOfTrainingData": 1013,
  "createdTime": "2010-01-01",
  "providers": [
    "Wuhan University"
  ],
  "keywords": [
    "Remote Sensing",
    "Scene Classification"
  ],
  "numberOfClasses": 19,
  "classes": [
    {
      "key": "Airport",
      "value": null
    },
    {
      "key": "Beach",
      "value": null
    },
    {
      "key": "Bridge",
      "value": null
    },
    {
      "key": "Commercial",
      "value": null
    },
    {
      "key": "Desert",
      "value": null
    },
    {
      "key": "Farmland",
      "value": null
    },
    {
      "key": "footballField",
      "value": null
    },
    {
      "key": "Forest",
      "value": null
    },
    {
      "key": "Industrial",
      "value": null
    },
    {
      "key": "Meadow",
      "value": null
    },
    {
      "key": "Mountain",
      "value": null
    },
    {
      "key": "Park",
      "value": null
    },
    {
      "key": "Parking",
      "value": null
    },
    {
      "key": "Pond",
      "value": null
    },
    {
      "key": "Port",
      "value": null
    },
    {
      "key": "railwayStation",
      "value": null
    },
    {
      "key": "Residential",
      "value": null
    },
    {
      "key": "River",
      "value": null
    },
    {
      "key": "Viaduct",
      "value": null
    }
  ],
  "extent": [
    -180,
    -90,
    180,
    90
  ],
  "referenceSystem": {
    "referenceSystemIdentifier": "EPSG:4326"
  },
  "tasks": [
    {
      "type": "AI_EOTask",
      "id": "whu_rs19-task",
      "description": "Structural high-resolution satellite image indexing",
      "taskType": "Scene Classification"
    }
  ],
  "data": [
    {
      "type": "AI_EOTrainingData",
      "id": "airport_01",
      "dataSources": [
        {
          "title": "googleEarth"
        }
      ],
      "dataURL": [
        "image/Airport/airport_01.jpg"
      ],
      "labels": [
        {
          "type": "AI_SceneLabel",
          "class": "Airport"
        }
      ]
    },
    …
  ]
}

{
  "doi": "10.1109/TGRS.2019.2917161",
  "algorithm": "FACNN",
  "metrics": [
    {
      "key": "Overall Accuracy",
      "value": 0.9881
    }
  ]
}

{
  "type": "AI_EOTrainingDataset",
  "id": "whu_rs19",
  "name": "WHU-RS19",
  "description": "Wuhan University-Remote Sensing 19 Categories (WHU-RS19) has 19 classes of remote sensing images scenes obtained from Google Earth",
  "license": "CC BY-SA 4.0",
  "amountOfTrainingData": 1013,
  "createdTime": "2010-01-01",
  "providers": [
    "Wuhan University"
  ],
  "keywords": [
    "Remote Sensing",
    "Scene Classification"
  ],
  "numberOfClasses": 19,
  "extent": [
    -180,
    -90,
    180,
    90
  ],
  "referenceSystem": {
    "referenceSystemIdentifier": "EPSG:4326"
  },
  "bands": [
    {
      "name": [
        {
          "code": "red"
        }
      ]
    },
    {
      "name": [
        {
          "code": "green"
        }
      ]
    },
    {
      "name": [
        {
          "code": "blue"
        }
      ]
    }
  ],
  "imageSize": "6000x7600",
  "classes": [
    {
      "key": "Airport",
      "value": null
    },
    {
      "key": "Beach",
      "value": null
    },
    {
      "key": "Bridge",
      "value": null
    },
    {
      "key": "Commercial",
      "value": null
    },
    {
      "key": "Desert",
      "value": null
    },
    {
      "key": "Farmland",
      "value": null
    },
    {
      "key": "footballField",
      "value": null
    },
    {
      "key": "Forest",
      "value": null
    },
    {
      "key": "Industrial",
      "value": null
    },
    {
      "key": "Meadow",
      "value": null
    },
    {
      "key": "Mountain",
      "value": null
    },
    {
      "key": "Park",
      "value": null
    },
    {
      "key": "Parking",
      "value": null
    },
    {
      "key": "Pond",
      "value": null
    },
    {
      "key": "Port",
      "value": null
    },
    {
      "key": "railwayStation",
      "value": null
    },
    {
      "key": "Residential",
      "value": null
    },
    {
      "key": "River",
      "value": null
    },
    {
      "key": "Viaduct",
      "value": null
    }
  ],
  "tasks": [
    {
      "type": "AI_EOTask",
      "id": "whu_rs19-task",
      "description": "Structural high-resolution satellite image indexing",
      "taskType": "Scene Classification"
    }
  ],
  "data": [
    {
      "type": "AI_EOTrainingData",
      "id": "airport_01",
      "dataSources": [
        {
          "title": "googleEarth"
        }
      ],
      "dataURL": [
        "image/Airport/airport_01.jpg"
      ],
      "labels": [
        {
          "type": "AI_SceneLabel",
          "class": "Airport"
        }
      ]
    },
    …
  ]
}

{
  "type": "AI_AbstractTrainingData",
  "id": "airport_01",
  "dataSources": [
    {
      "title": "googleEarth"
    }
  ],
  "extent": {
    "geographicElement": [
      {
        "westBoundLongitude": -171.76409,
        "eastBoundLongitude": -171.56578,
        "southBoundLatitude": -14.42443,
        "northBoundLatitude": -14.32568
      }
    ]
  },
  "dataTime": [
    "2002-05-30T09:30:10Z"
  ],
  "labels": [
    {
      "type": "AI_SceneLabel",
      "class": "Airport"
    }
  ]
}

 a)	"training"
 b)	"validation"
 c)	"test"
 d)	"retraining"

{
  "type": "AI_EOTrainingData",
  "id": "airport_01",
  "dataSources": [
    {
      "title": "googleEarth"
    }
  ],
  "extent": {
    "geographicElement": [
      {
        "westBoundLongitude": -171.76409,
        "eastBoundLongitude": -171.56578,
        "southBoundLatitude": -14.42443,
        "northBoundLatitude": -14.32568
      }
    ]
  },
  "dataTime": [
    "2002-05-30T09:30:10Z"
  ],
  "dataURL": [
    "image/Airport/airport_01.jpg"
  ],
  "spatialResolution": {
    "distance": {
      "value": "0.5",
      "uom": "m"
    }
  },
  "labels": [
    {
      "type": "AI_SceneLabel",
      "class": "Airport"
    }
  ]
}

{
  "type": "AI_AbstractTask",
  "id": "image-indexing-task",
  "description": "Structural high-resolution satellite image indexing"
}

{
  "type": "AI_EOTask",
  "id": "image-indexing-task",
  "description": "Structural high-resolution satellite image indexing",
  "taskType": "Scene Classification"
}

{
  "type": "AI_AbstractLabel",
  "isNegative": false,
  "confidence": 1.0
}

{
  "type": "AI_SceneLabel",
  "class": "Airport"
}

{
  "type": "AI_ObjectLabel",
  "class": "Truck",
  "object": {
    "type": "Feature",
    "properties": {
      "truncated": 0.0,
      "occluded": 0,
      "alpha": -1.57
    },
    "geometry": {
      "type": "Polygon",
      "coordinates": [
        [
          [
            2257.0,
            332.0
          ],
          [
            2271.0,
            332.0
          ],
          [
            2271.0,
            350.0
          ],
          [
            2257.0,
            350.0
          ],
          [
            2257.0,
            332.0
          ]
        ]
      ]
    }
  },
  "bboxType": "Horizontal BBox"
}

{
  "type": "AI_PixelLabel",
  "imageURL": [
    "/label_5classes/GF2_PMS1__L1A0000647767-MSS1_label.tif"
  ],
  "imageFormat": [
    "image/tiff; application=geotiff"
  ]
}

 a)	"image/tiff; application=geotiff"
 b)	"application/x-netcdf"
 c)	"image/png"
 d)	"image/jp2"

{
  "type": "AI_Labeling",
  "id": "0",
  "scope": {
    "level": "dataset",
    "levelDescription": [
      {
        "dataset": "whu_rs19"
      }
    ]
  },
  "labelingTime": "2015-04-20T07:04:15Z",
  "labelers": [
    {
      ..
    }
  ],
  "procedure": {
    ..
  }
}

{
  "type": "AI_Labeler",
  "id": "0",
  "name": "Tom"
}

{
  "type": "AI_LabelingProcedure",
  "id": "0",
  "methods": [
    "manual"
  ],
  "tools": [
    "ArcGIS"
  ]
}

 a)	"manual"
 b)	"semi-automatic"
 c)	"automatic"
 d)	"unknown"

{
  "type": "AI_ClassBalanceDegree",
  "measure": {
    "measureDescription": "Balance degree of label classes"
  },
  "evaluationMethod": [
    {
      "evaluationMethodDescription": "Counting the number of training samples belonging to each class and calculating the balance degree"
    }
  ],
  "result": [
    {
      "quantitativeResult": {
        "value": [
          93.5
        ],
        "valueUnit": "%"
      }
    }
  ]
}

{
  "type": "AI_TDChangeset",
  "id": "changeset-dota_v1.5",
  "datasetId": "dota_v1.5",
  "createdTime": "2019-01-01",
  "changeCount": 9,
  "modify": [
    {
      "type": "AI_EOTrainingData",
      "id": "P1228",
      "dataSources": [
        {
          "title": "GF"
        }
      ],
      "dataURL": [
        "train/images/P1228.png"
      ],
      "numberOfLabels": 50,
      "trainingType": "training",
      "labels": [
        {
          "type": "AI_ObjectLabel",
          "class": "ship",
          "object": {
            "type": "Feature",
            "properties": {},
            "geometry": {
              "type": "Polygon",
              "coordinates": [
                [
                  [
                    2306.0,
                    729.0
                  ],
                  [
                    2330.0,
                    729.0
                  ],
                  [
                    2330.0,
                    744.0
                  ],
                  [
                    2306.0,
                    744.0
                  ],
                  [
                    2306.0,
                    729.0
                  ]
                ]
              ]
            }
          },
          "bboxType": "Horizontal BBox"
        },
        …
      ]
    }
  ]
}

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).