Research map / Detection and segmentation / Tracking
XS-VID: A Large-Scale Benchmark for Small Object Detection and Tracking in Videos
Jiahao Guo, Ziyang Xu, Lianjun Wu, Fei Gao, Wenyu Liu, Xinggang Wang
Read the paperCheck 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 |
|---|---|---|---|
| add small-object techniques (upsampling, backbone, loss) (ablation) | YOLOv8 · XS-VID test | AP: 22.7 -> 24.5 | Improved |
| add Flow Mask Attention without mask supervision (ablation) | YOLOv8 + small-object techniques · XS-VID test | AP: 24.5 -> 24.3 | Hurt |
| add SAM2 mask auxiliary loss (ablation) | YOLOFT with FMA · XS-VID test | AP: 24.3 -> 25.2 | Improved |
| add video data augmentation (ablation) | YOLOFT · XS-VID test | AP: 25.2 -> 25.8 | Improved |
| use 3D conv instead of concatenation for temporal fusion (ablation) | FMA temporal fusion · XS-VID test | AP: 23.3 -> 24.7 | Improved |
| use ConvLSTM instead of 3D conv for temporal fusion (ablation) | FMA temporal fusion · XS-VID test | AP: 24.7 -> 23.5 | Hurt |
| add current-frame mask supervision (ablation) | YOLOFT · XS-VID test | AP: 23.2 -> 24.1 | Improved |
| add unweighted historical mask supervision (ablation) | YOLOFT with current mask · XS-VID test | AP: 24.1 -> 23.5 | Hurt |
| add half-weight historical mask supervision (ablation) | YOLOFT with current mask · XS-VID test | AP: 24.1 -> 24.6 | Improved |
| replace bilinear with ESPCN upsampling (ablation) | detector · XS-VID test | AP^s: 12.4 -> 13.7 | Improved |
| replace GIoU with NWD loss (ablation) | detector · XS-VID test | AP^s: 12.0 -> 12.9 | Improved |
| use 0.4*CIoU + 0.6*SAFit loss instead of GIoU (ablation) | detector · XS-VID test | AP^s: 12.0 -> 14.1 | Improved |
Closest papers by meaning
- TAO: A Large-Scale Benchmark for Tracking Any Object (ECCV 2020 · similarity 0.90)
- Towards Large-Scale Small Object Detection: Survey and Benchmarks (TPAMI 2023 · similarity 0.89)
- Event-based Tiny Object Detection: A Benchmark Dataset and Baseline (ICCV 2025 · similarity 0.89)
- TrackingNet: A Large-Scale Dataset and Benchmark for Object Tracking in the Wild (ECCV 2018 · similarity 0.89)
- OVT-B: A New Large-Scale Benchmark for Open-Vocabulary Multi-Object Tracking (NeurIPS 2024 · similarity 0.87)
- YOLOv10: Real-Time End-to-End Object Detection (NeurIPS 2024 · similarity 0.87)
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.
