Research map / Video / Video segmentation and tracking
Event-aided Dense and Continuous Point Tracking: Everywhere and Anytime
Zhexiong Wan, Jianqin Luo, Yuchao Dai, Gim Hee Lee
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 event data (ablation) | EDCPT · CVO third / DAVIS quarter | EPEall / AJ: 2.50 -> 2.23; 52.5 -> 55.4 | Improved |
| add image-to-trajectory consistency loss (ablation) | EDCPT with events · CVO third / DAVIS quarter | EPEall / AJ: 2.23 -> 2.20; 55.4 -> 55.8 | Improved |
| add event consistency loss (ablation) | EDCPT with Lic · CVO third / DAVIS quarter | EPEall / AJ: 2.20 -> 2.16; 55.8 -> 56.2 | Improved |
| add post aggregation instead of none (ablation) | EDCPT · CVO third / DAVIS quarter | EPEall / AJ: 2.54 -> 2.49; 51.9 -> 52.5 | Improved |
| replace post aggregation with feature-level streaming aggregation (ablation) | EDCPT · CVO third / DAVIS quarter | EPEall / AJ: 2.49 -> 2.42; 52.5 -> 53.3 | Improved |
| add start-point offset estimation (ablation) | EDCPT streaming · CVO third / DAVIS quarter | EPEall / AJ: 2.43 -> 2.42; 53.0 -> 53.3 | Improved |
| use SOLOFusion-style fusion instead of streaming recurrent fusion (ablation) | EDCPT · CVO third / DAVIS quarter | EPEall / AJ: 2.42 -> 2.45; 53.3 -> 52.7 | Hurt |
| use B-spline Nc=4 curve instead of linear (ablation) | EDCPT · CVO third / DAVIS quarter | EPEall / AJ: 2.42 -> 2.23; 53.3 -> 55.4 | Improved |
| increase control points from 4 to 5 (ablation) | EDCPT B-spline · CVO third / DAVIS quarter | EPEall / AJ: 2.23 -> 2.26; 55.4 -> 55.0 | No clear effect |
| use quadratic instead of linear motion (ablation) | EDCPT · CVO third / DAVIS quarter | EPEall / AJ: 2.42 -> 2.49; 53.3 -> 54.2 | Mixed |
| use EDCPT instead of DOT | CVO Extended · CVO (Extended) | EPEall: 5.07 -> 4.88 | Improved |
| use EDCPT instead of DOT with skipped frames | CVO Final third · CVO (Final) third | EPEall: 2.69 -> 2.16 | Improved |
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.
- FlowTrack — FlowTrack: Revisiting Optical Flow for Long-Range Dense Tracking (CVPR 2024) · better on 1 of 1 shared result
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
- TimeTracker: Event-based Continuous Point Tracking for Video Frame Interpolation with Non-linear Motion (CVPR 2025 · similarity 0.91)
- Online Dense Point Tracking with Streaming Memory (ICCV 2025 · similarity 0.90)
- Track-On: Transformer-based Online Point Tracking with Memory (ICLR 2025 · similarity 0.89)
- ETAP: Event-based Tracking of Any Point (CVPR 2025 · similarity 0.89)
- TrackIME: Enhanced Video Point Tracking via Instance Motion Estimation (NeurIPS 2024 · similarity 0.89)
- Tracktention: Leveraging Point Tracking to Attend Videos Faster and Better (CVPR 2025 · 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.
