Research map / Detection and segmentation / Tracking
End-To-End Representation Learning for Correlation Filter Based Tracking
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 |
|---|---|---|---|
| Add an end-to-end trained Correlation Filter layer to a shallow Siamese network | Siamese tracker · Validation set, depth one | tracking accuracy: 31% relative improvement | Helped |
| Add an end-to-end trained Correlation Filter layer to a shallow Siamese network | Siamese tracker · Validation set, depth two | tracking accuracy: 13% relative improvement | Helped |
| Add a Correlation Filter layer to sufficiently deep networks (ablation) | Siamese tracker · Validation set, depths three to five | tracking accuracy | No clear effect |
| Update the template with a running average instead of keeping it fixed (ablation) | Baseline and CFNet trackers · Validation set, all tested depths | tracking accuracy | Helped |
| Train features jointly with the Correlation Filter rather than apply CF after baseline training (ablation) | CFNet versus Baseline+CF · Validation set, depths one and two | tracking accuracy | Helped |
| Use baseline-trained features with a CF instead of end-to-end CFNet features (ablation) | CFNet · Validation set, deeper networks | tracking accuracy | No clear effect |
| Replace tracking-trained features with ImageNet classification features (ablation) | CF tracker · Validation set | tracking accuracy | Hurt |
| Make the Lagrange multipliers constant and learned offline rather than image-adaptive (ablation) | CFNet · Validation set | tracking accuracy | Hurt |
| Learn the desired response instead of using a fixed Gaussian response (ablation) | Correlation Filter layer · Tracking experiments | tracking accuracy | No clear effect |
Closest papers by meaning
- Correlation-Aware Deep Tracking (CVPR 2022 · similarity 0.88)
- ECO: Efficient Convolution Operators for Tracking (CVPR 2017 · similarity 0.88)
- Correlation Tracking via Joint Discrimination and Reliability Learning (CVPR 2018 · similarity 0.88)
- Robust Estimation of Similarity Transformation for Visual Object Tracking (AAAI 2019 · similarity 0.88)
- Attentional Correlation Filter Network for Adaptive Visual Tracking (CVPR 2017 · similarity 0.87)
- Context-Aware Deep Feature Compression for High-Speed Visual Tracking (CVPR 2018 · 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.