Research map / Detection and segmentation / Tracking
LaSOT: A High-quality Large-scale Single Object Tracking Benchmark
Heng Fan, Hexin Bai, Liting Lin, Fan Yang, Peng Chu, Ge Deng, Sijia Yu, Harshit, Mingzhen Huang, Juehuan Liu, Yong Xu, Chunyuan Liao and 2 more
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 |
|---|---|---|---|
| retrain on LaSOT full-overlap training split instead of ImageNet VID (ablation) | SiamFC · OTB-13 | success: 0.588 to 0.608 | Improved |
| retrain on LaSOT one-shot training split instead of ImageNet VID (ablation) | SiamFC · OTB-13 | success: 0.588 to 0.614 | Improved |
| retrain on LaSOT full-overlap training split instead of ImageNet VID (ablation) | SiamFC · OTB-15 | success: 0.565 to 0.582 | Improved |
| retrain on LaSOT training split instead of ImageNet VID (ablation) | CFNet · OTB-13 | success: 0.589 to 0.615 and 0.622 | Improved |
| retrain on LaSOT training split instead of ImageNet VID (ablation) | CFNet · OTB-15 | success: 0.568 to 0.593 and 0.598 | Improved |
| retrain on LaSOT training split instead of ImageNet VID (ablation) | SiamFC · LaSOT test (full overlap protocol) and one-shot | success: 0.336 to 0.342; 0.230 to 0.237 | Improved |
| retrain on LaSOT training split instead of ImageNet VID (ablation) | CFNet · LaSOT test (full overlap protocol) and one-shot | success: 0.275 to 0.286; 0.184 to 0.194 | Improved |
| use deeper backbone ResNet-50 instead of AlexNet (ablation) | SiamRPN++ · LaSOT test (full overlap protocol) | success: 0.433 to 0.495 | Improved |
| use deeper backbone ResNet-50 instead of AlexNet (ablation) | SiamRPN++ · LaSOT test (one-shot protocol) | success: 0.245 to 0.340 | Improved |
| use ResNet-50 instead of ResNet-18 (ablation) | DiMP · LaSOT test (full overlap protocol) | success: 0.534 to 0.560 | Improved |
| use ResNet-50 instead of ResNet-18 (ablation) | DiMP · LaSOT test (one-shot protocol) | success: 0.381 to 0.392 | Improved |
| add online model update (DiMP, LTMU) versus matching trackers without update | trackers · LaSOT full overlap | success: 0.517, 0.495, 0.467, 0.455 to 0.560, 0.539 | Improved |
Closest papers by meaning
- LaSOT: A High-Quality Benchmark for Large-Scale Single Object Tracking (CVPR 2019 · similarity 0.96)
- 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)
- VastTrack: Vast Category Visual Object Tracking (NeurIPS 2024 · similarity 0.91)
- OVT-B: A New Large-Scale Benchmark for Open-Vocabulary Multi-Object Tracking (NeurIPS 2024 · similarity 0.89)
- DiffusionTrack: Diffusion Model for Multi-Object Tracking (AAAI 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.
