Research map / Detection and segmentation / Open-set and few-shot detection
YouTube-BoundingBoxes: A Large High-Precision Human-Annotated Data Set for Object Detection in Video
Esteban Real, Jonathon Shlens, Stefano Mazzocchi, Xin Pan, Vincent Vanhoucke
Check your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Train the image classifier on YT-BB rather than COCO | Inception-v3 image classification · YT-BB | mAP: 0.77 -> 0.93 | Improved |
| Train the detector on YT-BB rather than COCO | Faster-RCNN object detection · YT-BB | mAP: 0.37 -> 0.59 | Improved |
| Evaluate the COCO-trained classifier on YT-BB instead of COCO | Inception-v3 image classification · COCO-trained model evaluated on COCO versus YT-BB | mAP: 0.83 -> 0.77 | Hurt |
| Evaluate the COCO-trained detector on YT-BB instead of COCO | Faster-RCNN object detection · COCO-trained model evaluated on COCO versus YT-BB | mAP: 0.43 -> 0.37 | Hurt |
| Evaluate the YT-BB-trained classifier on COCO instead of YT-BB | Inception-v3 image classification · YT-BB-trained model evaluated on YT-BB versus COCO | mAP: 0.93 -> 0.66 | Hurt |
| Evaluate the YT-BB-trained detector on COCO instead of YT-BB | Faster-RCNN object detection · YT-BB-trained model evaluated on YT-BB versus COCO | mAP: 0.59 -> 0.31 | Hurt |
| Average predictions across each video segment (ablation) | Inception-v3 trained on YT-BB · YT-BB | mAP: 0.93 -> 0.95 | No clear effect |
| Down-weight weak detections that do not overlap detections in neighboring frames (ablation) | Faster-RCNN trained on COCO · YT-BB | mAP: 0.37 -> 0.36 | Mixed |
Closest papers by meaning
- YouTube-VOS: Sequence-to-Sequence Video Object Segmentation (ECCV 2018 · similarity 0.87)
- Open-Vocabulary Object Detection Using Captions (CVPR 2021 · similarity 0.87)
- Generating Masks From Boxes by Mining Spatio-Temporal Consistencies in Videos (ICCV 2021 · similarity 0.87)
- Detecting 11K Classes: Large Scale Object Detection Without Fine-Grained Bounding Boxes (ICCV 2019 · similarity 0.87)
- We Don't Need No Bounding-Boxes: Training Object Class Detectors Using Only Human Verification (CVPR 2016 · similarity 0.87)
- Object Detection Benchmarks are Incomplete: The Role of Label Errors and Annotation Uncertainty (NeurIPS 2026 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
