Research map / Video / Video segmentation and tracking
MATE: Motion-Augmented Temporal Consistency for Event-based Point Tracking
Han Han, Wei Zhai, Yang Cao, Bin Li, Zheng-jun Zha
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 MGM plane-fitting kinematic vectors to appearance-only matching (ablation) | MATE · Ev-PointOdyssey | MTE: 43.64 -> 38.72 | Improved |
| add plane fitting alone (no MLP correction) on Survival_50 (ablation) | MATE · Ev-PointOdyssey | Survival_50: 0.488 -> 0.485 | No clear effect |
| add MLP-based correction of kinematic vectors (ablation) | MATE MGM · Ev-PointOdyssey | Survival_50: 0.485 -> 0.513 | Improved |
| add cross-attention with kinematic guidance in VMA (ablation) | MATE · Ev-PointOdyssey | Survival_50: 0.513 -> 0.538 | Improved |
| add 1D temporal convolution (short-term path) to VMA (ablation) | MATE · Ev-PointOdyssey | delta_avg: 0.360 -> 0.358 | No clear effect |
| add temporal attention (long-term path) to VMA (ablation) | MATE · Ev-PointOdyssey | MTE: 27.63 -> 25.63 | Improved |
| use time surface instead of event image as input representation (ablation) | MATE baseline without VMA · Ev-PointOdyssey (192x256) | Survival_50: 0.443 -> 0.477 | Improved |
| use time surface instead of voxel grid (ablation) | MATE baseline without VMA · Ev-PointOdyssey (192x256) | Survival_50: 0.457 -> 0.477 | Improved |
| event-only MATE versus video-based Cotracker3 | MATE · Ev-PointOdyssey | Survival_50: 0.409 -> 0.560 | Improved |
| apply video trackers directly to event data instead of video | Cotracker3 · Ev-PointOdyssey | Survival_50: 0.409 -> 0.428 | Mixed |
| gamma loss weight set to 0.8 and 6 iterations (ablation) | MATE · Ev-PointOdyssey | Survival_50 | Improved |
Closest papers by meaning
- ETAP: Event-based Tracking of Any Point (CVPR 2025 · similarity 0.93)
- Event-aided Dense and Continuous Point Tracking: Everywhere and Anytime (ICCV 2025 · similarity 0.90)
- Match-Any-Events: Zero-Shot Motion-Robust Feature Matching Across Wide Baselines for Event Cameras (ECCV 2026 · similarity 0.89)
- TTAPFormer: Robust Arbitrary Point Tracking via Transient Asynchronous Fusion of Frames and Events (CVPR 2026 · similarity 0.89)
- TimeTracker: Event-based Continuous Point Tracking for Video Frame Interpolation with Non-linear Motion (CVPR 2025 · similarity 0.88)
- Track-On: Transformer-based Online Point Tracking with Memory (ICLR 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.
