Research map / Architectures and efficiency / Pruning and efficient tuning
Context-Aware Deep Feature Compression for High-Speed Visual Tracking
Read the paperCheck 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 |
|---|---|---|---|
| add initial adaptation of the selected expert auto-encoder (ablation) | TRACA · CVPR2013 | precision: 83.2% -> 89.8% | Helped |
| add the correlation-filter orthogonality loss during initial adaptation (ablation) | TRACA · CVPR2013 | precision: 82.2% -> 89.8% | Helped |
| add extrinsic denoising during expert auto-encoder pre-training (ablation) | TRACA · CVPR2013 | precision: 75.4% -> 89.8% | Helped |
| use dimension corruption rather than dropout, retaining feature-vector exchange (ablation) | TRACA · CVPR2013 | precision: 78.5% -> 89.8% | Helped |
| select the expert auto-encoder with the context-aware network rather than randomly (ablation) | TRACA · CVPR2013 | precision: 84.4% -> 89.8% | Helped |
| select the expert auto-encoder with the context-aware network rather than by smallest L2 generation error (ablation) | TRACA · CVPR2013 | precision: 87.7% -> 89.8% | Helped |
| cluster training samples by compressed feature space rather than VOC2012 class labels (ablation) | TRACA · CVPR2013 | precision: 76.6% -> 89.8% | Helped |
| add multi-stage reconstruction distance rather than using only final-output Euclidean distance (ablation) | TRACA · CVPR2013 | precision: 76.4% -> 89.8% | Helped |
| use multiple context-specialized expert auto-encoders rather than a single base auto-encoder (ablation) | TRACA · CVPR2013 | precision: 76.2% -> 89.8% | Helped |
Closest papers by meaning
- Context-Aware Correlation Filter Tracking (CVPR 2017 · similarity 0.89)
- Correlation-Aware Deep Tracking (CVPR 2022 · similarity 0.88)
- Multi-Cue Correlation Filters for Robust Visual Tracking (CVPR 2018 · similarity 0.88)
- Target-Aware Deep Tracking (CVPR 2019 · similarity 0.87)
- Less Is More: Token Context-Aware Learning for Object Tracking (AAAI 2025 · similarity 0.86)
- Efficient Motion Prompt Learning for Robust Visual Tracking (ICML 2025 · similarity 0.85)
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.