Research map / Label-efficient and robust learning / Self-supervised and contrastive
MAST: A Memory-Augmented Self-Supervised Tracker
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use MAST instead of CorrFlow | Self-supervised dense tracking · DAVIS-2017 validation | J&F (Mean): 50.3 vs 65.5 | Helped |
| Use MAST instead of CorrFlow | Self-supervised dense tracking · YouTube-VOS validation | Overall: 46.6 -> 64.2 | Helped |
| Use Lab colorspace rather than RGB (ablation) | MAST · DAVIS-2017 validation | J (Mean): 52.7 -> 61.2 | Helped |
| Use Lab colorspace rather than HSV (ablation) | MAST · DAVIS-2017 validation | J (Mean): 54.3 -> 61.2 | Helped |
| Use regression loss instead of classification loss on Lab colors (ablation) | MAST · DAVIS-2017 validation | J (Mean): 47.1 -> 61.2 | Helped |
| Use single-color-space matching instead of cross-color-space matching (ablation) | MAST · DAVIS-2017 validation | J (Mean): 46.8 -> 61.2 | Helped |
| Add image-feature alignment (ablation) | MAST · DAVIS-2017 validation | J (Mean): +2.1 | Helped |
| Add short-term memory alongside long-term memory (ablation) | MAST · DAVIS-2017 validation | J (Mean): 44.6 -> 61.2 | Helped |
| Add long-term memory alongside short-term memory (ablation) | MAST · DAVIS-2017 validation | J (Mean): 57.3 -> 61.2 | Helped |
| Quantize pixel class probabilities during propagation (ablation) | MAST · DAVIS-2017 validation | J (Mean): +4.2 | Helped |
| Train on YouTube-VOS raw videos instead of OxUvA (ablation) | MAST · DAVIS-2017 validation | J (Mean): 61.2 -> 63.3 | Helped |
| Train on ImageNet VID instead of OxUvA (ablation) | MAST · DAVIS-2017 validation | J (Mean): 61.2 -> 60.0 | Hurt |
Closest papers by meaning
- A Distractor-Aware Memory for Visual Object Tracking with SAM2 (CVPR 2025 · similarity 0.89)
- Video Diffusion Models Excel at Tracking Similar-Looking Objects Without Supervision (NeurIPS 2025 · similarity 0.89)
- TrackIME: Enhanced Video Point Tracking via Instance Motion Estimation (NeurIPS 2024 · similarity 0.88)
- Track-On: Transformer-based Online Point Tracking with Memory (ICLR 2025 · similarity 0.88)
- TrackCraft3R: Repurposing Video Diffusion Transformers for Dense 3D Tracking (NeurIPS 2026 · similarity 0.87)
- VastTrack: Vast Category Visual Object Tracking (NeurIPS 2024 · 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.