vis4d.zoo.base.datasets.coco

COCO dataset config.

get_coco_detection_cfg(data_root='data/coco', train_split='train2017', train_keys_to_load=('images', 'boxes2d', 'boxes2d_classes'), train_cached_file_path='data/coco/train.pkl', test_split='val2017', test_keys_to_load=('images', 'original_images', 'boxes2d', 'boxes2d_classes'), test_cached_file_path='data/coco/val.pkl', cache_as_binary=True, data_backend=None, image_size=(800, 1333), samples_per_gpu=2, workers_per_gpu=2)

Get the default config for COCO detection.

Return type:

DataConfig

get_coco_sem_seg_cfg(data_root='data/coco', train_split='train2017', train_keys_to_load=('images', 'seg_masks'), test_split='val2017', test_keys_to_load=('images', 'seg_masks'), data_backend=None, image_size=(520, 520), samples_per_gpu=2, workers_per_gpu=2)

Get the default config for COCO semantic segmentation.

Return type:

DataConfig

Modules

vis4d.zoo.base.datasets.coco.detection	COCO data loading config for object detection.
vis4d.zoo.base.datasets.coco.sem_seg	COCO data loading config for for semantic segmentation.