Research map / Detection and segmentation / Tracking
GOT-10k: A Large High-Diversity Benchmark for Generic Object Tracking in the Wild
Lianghua Huang, Xin Zhao, Kaiqi Huang
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 |
|---|---|---|---|
| train on GOT-10k instead of ImageNet-VID (ablation) | MemTracker · GOT-10k test | mAO: 0.447 -> 0.460 | Improved |
| train on GOT-10k instead of ImageNet-VID (ablation) | SiamFCv2 · OTB2015 | score: 0.613 -> 0.621 | Improved |
| train on GOT-10k instead of ImageNet-VID (ablation) | MDNet · OTB2015 | score: 0.673 -> 0.637 | Hurt |
| train on GOT-10k instead of ImageNet-VID (ablation) | GOTURN · OTB2015 | score: 0.427 -> 0.413 | Hurt |
| train on class-balanced rather than imbalanced data (ablation) | MemTracker · GOT-10k (100-class subset) | mAO: 0.412 -> 0.431 | Improved |
| train on class-balanced rather than imbalanced data (ablation) | MDNet · GOT-10k (100-class subset) | mAO: 0.340 -> 0.357 | Improved |
| train on class-balanced rather than imbalanced data (ablation) | GOTURN · GOT-10k (100-class subset) | mAO: 0.421 -> 0.408 | Hurt |
| train on class-balanced rather than imbalanced data (ablation) | SiamFCv2 · GOT-10k (100-class subset) | mAO: 0.407 -> 0.399 | Hurt |
| test on seen instead of unseen classes (ablation) | MemTracker · GOT-10k (3 random repeats) | mAO: 0.588/0.567/0.598 -> 0.595/0.589/0.612 | Improved |
| test on seen instead of unseen classes (ablation) | MDNet · GOT-10k (3 random repeats) | mAO: 0.439/0.441/0.449 -> 0.443/0.437/0.450 | No clear effect |
| use balanced instead of imbalanced test data (ablation) | 25 trackers · GOT-10k test sampling | std. of ranks (mAO): 1.213 -> 1.102 | Improved |
| use balanced instead of imbalanced test data (ablation) | 25 trackers · GOT-10k test sampling | std. of ranks (mSR): 1.241 -> 1.114 | Improved |
Closest papers by meaning
- TrackingNet: A Large-Scale Dataset and Benchmark for Object Tracking in the Wild (ECCV 2018 · similarity 0.93)
- TAO: A Large-Scale Benchmark for Tracking Any Object (ECCV 2020 · similarity 0.92)
- Long-term Tracking in the Wild: a Benchmark (ECCV 2018 · similarity 0.92)
- VastTrack: Vast Category Visual Object Tracking (NeurIPS 2024 · similarity 0.90)
- NetTrack: Tracking Highly Dynamic Objects with a Net (CVPR 2024 · similarity 0.88)
- OVT-B: A New Large-Scale Benchmark for Open-Vocabulary Multi-Object Tracking (NeurIPS 2024 · similarity 0.88)
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.
