📖 ML-STAC Collection Schema 📜

{
    "$defs": {
        "Author": {
            "properties": {
                "name": {
                    "title": "Name",
                    "type": "string"
                },
                "orcid": {
                    "anyOf": [
                        {
                            "type": "string"
                        },
                        {
                            "type": "null"
                        }
                    ],
                    "default": null,
                    "title": "Orcid"
                },
                "organization": {
                    "anyOf": [
                        {
                            "type": "string"
                        },
                        {
                            "type": "null"
                        }
                    ],
                    "default": null,
                    "title": "Organization"
                }
            },
            "required": [
                "name"
            ],
            "title": "Author",
            "type": "object"
        },
        "Authors": {
            "properties": {
                "authors": {
                    "items": {
                        "$ref": "#/$defs/Author"
                    },
                    "title": "Authors",
                    "type": "array"
                },
                "additional_comments": {
                    "anyOf": [
                        {
                            "type": "string"
                        },
                        {
                            "type": "null"
                        }
                    ],
                    "default": null,
                    "title": "Additional Comments"
                },
                "curated_by": {
                    "$ref": "#/$defs/HyperLink"
                },
                "reference": {
                    "$ref": "#/$defs/HyperLink"
                }
            },
            "required": [
                "authors",
                "curated_by",
                "reference"
            ],
            "title": "Authors",
            "type": "object"
        },
        "Catalog": {
            "additionalProperties": false,
            "description": "ML-STAC Catalog dataclass.",
            "properties": {
                "name": {
                    "enum": [
                        "train",
                        "test",
                        "val"
                    ],
                    "title": "Name",
                    "type": "string"
                },
                "n_items": {
                    "title": "N Items",
                    "type": "integer"
                }
            },
            "required": [
                "name",
                "n_items"
            ],
            "title": "Catalog",
            "type": "object"
        },
        "Extent": {
            "properties": {
                "spatial": {
                    "$ref": "#/$defs/SpatialExtent"
                },
                "temporal": {
                    "$ref": "#/$defs/TemporalExtent"
                }
            },
            "required": [
                "spatial",
                "temporal"
            ],
            "title": "Extent",
            "type": "object"
        },
        "HyperLink": {
            "properties": {
                "link": {
                    "title": "Link",
                    "type": "string"
                }
            },
            "required": [
                "link"
            ],
            "title": "HyperLink",
            "type": "object"
        },
        "License": {
            "properties": {
                "name": {
                    "title": "Name",
                    "type": "string"
                },
                "link": {
                    "$ref": "#/$defs/HyperLink"
                }
            },
            "required": [
                "name",
                "link"
            ],
            "title": "License",
            "type": "object"
        },
        "Licenses": {
            "additionalProperties": false,
            "properties": {
                "licenses": {
                    "items": {
                        "$ref": "#/$defs/License"
                    },
                    "title": "Licenses",
                    "type": "array"
                },
                "additional_comments": {
                    "anyOf": [
                        {
                            "type": "string"
                        },
                        {
                            "type": "null"
                        }
                    ],
                    "default": null,
                    "title": "Additional Comments"
                }
            },
            "required": [
                "licenses"
            ],
            "title": "Licenses",
            "type": "object"
        },
        "Reviewer": {
            "properties": {
                "name": {
                    "title": "Name",
                    "type": "string"
                },
                "orcid": {
                    "anyOf": [
                        {
                            "type": "string"
                        },
                        {
                            "type": "null"
                        }
                    ],
                    "default": null,
                    "title": "Orcid"
                },
                "score": {
                    "enum": [
                        0,
                        1,
                        2,
                        3,
                        4,
                        5
                    ],
                    "title": "Score",
                    "type": "integer"
                },
                "issue_link": {
                    "$ref": "#/$defs/HyperLink"
                }
            },
            "required": [
                "name",
                "score",
                "issue_link"
            ],
            "title": "Reviewer",
            "type": "object"
        },
        "Reviewers": {
            "properties": {
                "reviewers": {
                    "anyOf": [
                        {
                            "items": {
                                "$ref": "#/$defs/Reviewer"
                            },
                            "type": "array"
                        },
                        {
                            "type": "null"
                        }
                    ],
                    "default": null,
                    "title": "Reviewers"
                }
            },
            "title": "Reviewers",
            "type": "object"
        },
        "SpatialExtent": {
            "properties": {
                "bbox": {
                    "items": {
                        "type": "number"
                    },
                    "title": "Bbox",
                    "type": "array"
                }
            },
            "required": [
                "bbox"
            ],
            "title": "SpatialExtent",
            "type": "object"
        },
        "TemporalExtent": {
            "properties": {
                "interval": {
                    "items": {
                        "format": "date-time",
                        "type": "string"
                    },
                    "title": "Interval",
                    "type": "array"
                }
            },
            "required": [
                "interval"
            ],
            "title": "TemporalExtent",
            "type": "object"
        }
    },
    "additionalProperties": true,
    "description": "ML-STAC Catalog dataclass.",
    "properties": {
        "mlstac_version": {
            "title": "Mlstac Version",
            "type": "string"
        },
        "name": {
            "title": "Name",
            "type": "string"
        },
        "train": {
            "$ref": "#/$defs/Catalog"
        },
        "val": {
            "$ref": "#/$defs/Catalog"
        },
        "test": {
            "$ref": "#/$defs/Catalog"
        },
        "authors": {
            "$ref": "#/$defs/Authors"
        },
        "licenses": {
            "$ref": "#/$defs/Licenses"
        },
        "split_strategy": {
            "enum": [
                "random sampling",
                "spatial stratified random sampling",
                "spatiotemporal stratified random sampling",
                "other"
            ],
            "title": "Split Strategy",
            "type": "string"
        },
        "cv_task": {
            "enum": [
                "multi-class classification",
                "binary classification",
                "object detection",
                "multi-class semantic segmentation",
                "binary semantic segmentation",
                "instance segmentation",
                "image to image",
                "multi-image to image",
                "image to multi-image",
                "multi-image to multi-image",
                "regression",
                "multi-task",
                "other"
            ],
            "title": "Cv Task",
            "type": "string"
        },
        "sensor": {
            "enum": [
                "sentinel-2 msi",
                "landsat-8 oli/tirs",
                "landsat-8 oli",
                "landsat-8 tirs",
                "landsat-7 etm+",
                "landsat-7 etm+ only slc-off",
                "landsat-5 tm",
                "landsat-4 tm",
                "landsat-1 to 5 mss",
                "modis",
                "modis-derived",
                "sentinel-1 sar",
                "aerial",
                "uav",
                "aster",
                "hyperion",
                "cbers",
                "planetscope",
                "rapideye",
                "spot 1 to 5",
                "spot 6/7",
                "ikonos",
                "worldview-2",
                "worldview-3",
                "other"
            ],
            "title": "Sensor",
            "type": "string"
        },
        "bands": {
            "items": {
                "enum": [
                    "aerosol",
                    "blue",
                    "green1",
                    "green",
                    "yellow",
                    "red",
                    "rededge1",
                    "rededge2",
                    "rededge3",
                    "nir",
                    "nir2",
                    "watervapor",
                    "cirrus",
                    "swir1",
                    "swir2",
                    "thermal1",
                    "thermal2",
                    "hv",
                    "hh",
                    "vv",
                    "vh",
                    "sar-derived",
                    "dem",
                    "dem-derived",
                    "landcover",
                    "landcover-derived",
                    "other"
                ],
                "type": "string"
            },
            "title": "Bands",
            "type": "array"
        },
        "review": {
            "$ref": "#/$defs/Reviewers"
        },
        "extent": {
            "$ref": "#/$defs/Extent"
        },
        "dtype_input": {
            "enum": [
                "bool",
                "uint8",
                "int8",
                "uint16",
                "int16",
                "int32",
                "float16",
                "float32",
                "float64"
            ],
            "title": "Dtype Input",
            "type": "string"
        },
        "dtype_target": {
            "enum": [
                "bool",
                "uint8",
                "int8",
                "uint16",
                "int16",
                "int32",
                "float16",
                "float32",
                "float64"
            ],
            "title": "Dtype Target",
            "type": "string"
        },
        "dtype_reference": {
            "enum": [
                "bool",
                "uint8",
                "int8",
                "uint16",
                "int16",
                "int32",
                "float16",
                "float32",
                "float64"
            ],
            "title": "Dtype Reference",
            "type": "string"
        },
        "size": {
            "title": "Size",
            "type": "integer"
        },
        "keywords": {
            "items": {
                "type": "string"
            },
            "title": "Keywords",
            "type": "array"
        },
        "description": {
            "title": "Description",
            "type": "string"
        },
        "extra": {
            "additionalProperties": {
                "type": "string"
            },
            "title": "Extra",
            "type": "object"
        },
        "issue_link": {
            "$ref": "#/$defs/HyperLink"
        }
    },
    "required": [
        "mlstac_version",
        "name",
        "train",
        "val",
        "test",
        "authors",
        "licenses",
        "split_strategy",
        "cv_task",
        "sensor",
        "bands",
        "review",
        "extent",
        "dtype_input",
        "dtype_target",
        "dtype_reference",
        "size",
        "keywords",
        "description",
        "extra",
        "issue_link"
    ],
    "title": "Collection",
    "type": "object"
}