| configs: | |
| - config_name: detection | |
| data_files: | |
| - split: train | |
| path: "detection/dave_vru_detection_annotations_train.json" | |
| - split: val | |
| path: "detection/dave_vru_detection_annotations_val.json" | |
| - config_name: retrieval | |
| data_files: | |
| - split: train | |
| path: "moment_retrieval/train_mr.jsonl" | |
| - split: test | |
| path: "moment_retrieval/test_mr.jsonl" | |
| - config_name: recognition | |
| data_files: | |
| - split: train | |
| path: "multi_label_video_action_recognition/train_ml.csv" | |
| - split: val | |
| path: "multi_label_video_action_recognition/val_ml.csv" | |
| - config_name: localization | |
| data_files: | |
| - split: train | |
| path: "spatio_temporal_action_localization/dave_spatio_temporal_train.json" | |
| - split: val | |
| path: "spatio_temporal_action_localization/dave_spatio_temporal_val.json" | |
| - config_name: tracking | |
| data_files: | |
| - split: train | |
| path: "tracking/dave_tracking_train.json" | |
| - split: val | |
| path: "tracking/dave_tracking_val.json" | |
| DAVE can benchmark video tasks like Tracking, Detection, Spatiotemporal Action Localization, Language-Visual Moment retrieval, and Multi-label Video Action Recognition. We offer all the structured annotations. This is for review purposes since we didn't release the paper by the deadline of this submission. | |
| --- | |
| license: apache-2.0 | |
| task_categories: | |
| - video-classification | |
| language: | |
| - en | |
| pretty_name: dave | |
| size_categories: | |
| - 10M<n<100M | |
| --- |