Supported Data Formats ###################### .. toctree:: :maxdepth: 1 :hidden: ade20k2017 ade20k2020 align_celeba arrow ava_action brats brats_numpy celeba cifar cityscapes coco common_semantic_segmentation common_super_resolution cvat datumaro_binary datumaro dota icdar image_zip imagenet kaggle kinetics kitti kitti_raw labelme lfw mapillary_vistas market1501 mars mmdet mnist mot mots mpii_json mpii mvtec nyu_depth_v2 open_images pascal_voc roboflow segment_anything sly_pointcloud synthia tabular vgg_face2 video vott_csv vott_json wider_face yolo yolo_ultralytics * ADE20k (v2017) (import-only) * `Format specification `_ * `Dataset example `_ * `Format documentation <./ade20k2017.md>`_ * ADE20k (v2020) (import-only) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * Align CelebA (``classification``, ``landmarks``) (import-only) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * BraTS (``segmentation``) (import-only) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * BraTS Numpy (``detection``, ``segmentation``) (import-only) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * CamVid (``segmentation``) * `Format specification `_ * `Dataset example `_ * CelebA (``classification``, ``detection``, ``landmarks``) (import-only) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * CIFAR-10/100 (``classification``) * `Format specification `_ * `Dataset example CIFAR-10 `_ * `Dataset example CIFAR-100 `_ * `Format documentation `_ * Cityscapes (``segmentation``) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * Common Semantic Segmentation (``segmentation``) (import-only) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * Common Super Resolution * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * CVAT (`for images`, `for video` (import-only)) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * DOTA (``detection_rotated``) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * ICDAR13/15 (``word recognition``, ``text localization``, ``text segmentation``) * `Format specification `_ * `Dataset example `_ * ImageNet (``classification``, ``detection``) * `Dataset example `_ * `Dataset example (txt for classification) `_ * Detection format is the same as in PASCAL VOC * `Format documentation `_ * Kaggle (``classification``, ``detection``, ``segmentation``) (import-only) * `Dataset examples `_ * `Format documentation `_ * KITTI (``segmentation``, ``detection``) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * KITTI 3D (``raw``, ``tracklets``, ``velodyne points``) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * Kinetics 400/600/700 * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * LabelMe (``labels``, ``boxes``, ``masks``) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * LFW (``classification``, ``person re-identification``, ``landmarks``) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * Mapillary Vistas (``segmentation``) (import-only) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * Market-1501 (``person re-identification``) * `Format specification `_ * `Dataset example `_ * MARS (import-only) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * MMDet-COCO (``detection``, ``segmentation``) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * MNIST (``classification``) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * MNIST in CSV (``classification``) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * MOT sequences * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * MOTS (png) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * MPII Human Pose (``detection``, ``pose estimation``) (import-only) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * MPII Human Pose JSON (``detection``, ``pose estimation``) (import-only) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * MS COCO (``image info``, ``instances``, ``person keypoints``, ``captions``, ``labels``, ``panoptic``, ``stuff``) * `Format specification `_ * `Dataset example `_ * ``labels`` are our extension - like `instances` with only `category_id` * `Format documentation `_ * Roboflow (import-only) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * NYU Depth Dataset V2 (``depth estimation``) (import-only) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * OpenImages (``classification``, ``detection``, ``segmentation``) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * PASCAL VOC (``classification``, ``detection``, ``segmentation``, ``action classification``, ``person layout``) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * Segment Anything (a.k.a SA-1B) (``detection``, ``segmentation``) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * Supervisely (``pointcloud``) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * SYNTHIA (``segmentation``) (import-only) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * Tabular (``classification``, ``regression``) (import/export only) * `Dataset example `_ * `Format documentation `_ * TF Detection API (``bboxes``, ``masks``) * Format specifications: `[bboxes] `_, `[masks] `_ * `Dataset example `_ * VGGFace2 (``landmarks``, ``bboxes``) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * VoTT CSV (``detection``) (import-only) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * VoTT JSON (``detection``) (import-only) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * WIDERFace (``bboxes``) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * YOLO (``bboxes``) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ * YOLO-Ultralytics (``bboxes``) * `Format specification `_ * `Dataset example `_ * `Format documentation `_ Supported Annotation Types ########################## * Labels * Bounding Boxes * Polygons * Polylines * (Segmentation) Masks * (Key-) Points * Captions * 3D cuboids * Super Resolution Annotation * Depth Annotation * Ellipses * Hash Keys Datumaro does not separate datasets by tasks like classification, detection, segmentation, etc. Instead, datasets can have any annotations. When a dataset is exported in a specific format, only relevant annotations are exported. Dataset Meta Info File ###################### It is possible to use classes that are not original to the format. To do this, use ``dataset_meta.json``. .. code-block:: json { "label_map": {"0": "background", "1": "car", "2": "person"}, "segmentation_colors": [[0, 0, 0], [255, 0, 0], [0, 0, 255]], "background_label": "0" } - ``label_map`` is a dictionary where the class ID is the key and the class name is the value. - ``segmentation_colors`` is a list of channel-wise values for each class. This is only necessary for the segmentation task. - ``background_label`` is a background label ID in the dataset.