Research map / Detection and segmentation / Tracking
Towards Real-World Visual Tracking With Temporal Contexts
Ziang Cao, Ziyuan Huang, Liang Pan, Shiwei Zhang, Ziwei Liu, Changhong Fu
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 temporal information filter to AT-Trans (single-frame training) (ablation) | Transformer · UAV123 overall | success: 0.550 -> 0.573 | Improved |
| add temporal information filter to AT-Trans (multi-frame training) (ablation) | Transformer · UAV123 overall | success: 0.508 -> 0.604 | Improved |
| multi-frame training without filtering invalid information (ablation) | Transformer · UAV123 overall | success: 0.550 -> 0.508 | Hurt |
| use convolutional (first-frame) init instead of random init of temporal prior (ablation) | AT-Trans · UAV123 overall | success: 0.586 -> 0.604 | Improved |
| use previous-frame refined map as query instead of current feature map F_t (ablation) | AT-Trans · UAV123 overall | success: 0.580 -> 0.604 | Improved |
| increase training video length from 1 to 4 with online TAdaConv+AT-Trans (ablation) | TCTrack · OTB100 offline | success: 0.601 -> 0.624 | Improved |
| replace online TAdaConv with attention-based ATT-TAdaConv (ablation) | TCTrack · OTB100 offline | success: 0.624 -> 0.642 | Improved |
| add curriculum learning (video length 2->4) (ablation) | ATT-TAdaConv+AT-Trans · GOT-10K online | AO: 0.333 -> 0.375 | Improved |
| add curriculum learning (video length 2->4) (ablation) | ATT-TAdaConv+AT-Trans · OTB100 online | success: 0.510 -> 0.543 | Improved |
| add feature-level temporal context with video length 2 or 3 (ablation) | Online TAdaConv+AT-Trans · GOT-10K online | AO: 0.281 -> 0.274 | Hurt |
| add temporal modeling at both feature and similarity levels (full TCTrack++) (ablation) | spatial backbone baseline · GOT-10K online | AO: 0.278 -> 0.375 | Improved |
| use ResNet50 instead of AlexNet backbone (ablation) | TCTrack++ · UAV123 | online success: 0.519 -> 0.472 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- TCTrack — TCTrack: Temporal Contexts for Aerial Tracking (CVPR 2022) · better on 6 of 6 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- ODTrack: Online Dense Temporal Token Learning for Visual Tracking (AAAI 2024 · similarity 0.87)
- Emergent Temporal Correspondences from Video Diffusion Transformers (NeurIPS 2025 · similarity 0.87)
- Temporal Adaptive RGBT Tracking with Modality Prompt (AAAI 2024 · similarity 0.87)
- HIPTrack: Visual Tracking with Historical Prompts (CVPR 2024 · similarity 0.87)
- VastTrack: Vast Category Visual Object Tracking (NeurIPS 2024 · similarity 0.87)
- Exploiting Multimodal Spatial-temporal Patterns for Video Object Tracking (AAAI 2025 · 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.
