Research map / Architectures and efficiency / CNN design
ECO: Efficient Convolution Operators for 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 factorized convolution to C-COT (ablation) | C-COT · VOT2016 | EAO: 0.331 -> 0.342 | Helped |
| Add the compact sample space model (ablation) | C-COT with factorized convolution · VOT2016 | EAO: 0.342 -> 0.352 | Helped |
| Add sparse filter updates every sixth frame (ablation) | C-COT with factorized convolution and sample space model · VOT2016 | EAO: 0.352 -> 0.374 | Helped |
| Add factorized convolution (ablation) | C-COT · VOT2016 | FPS: 0.3 -> 1.1 | Helped |
| Add the compact sample space model (ablation) | C-COT with factorized convolution · VOT2016 | FPS: 1.1 -> 2.6 | Helped |
| Update the filter every sixth frame rather than every frame (ablation) | C-COT with factorized convolution and sample space model · VOT2016 | FPS: 2.6 -> 6.0 | Helped |
| Replace discriminatively learned projection with PCA (ablation) | Factorized convolution · VOT2016 | EAO: 0.331 to 0.319 | Hurt |
| Reduce the standard C-COT training sample set from 400 to 50 samples (ablation) | C-COT · VOT2016 | EAO: 0.342 to 0.338 (- 1.2 %) | Hurt |
| Represent training samples with 50 GMM components (ablation) | ECO sample space model · VOT2016 | EAO: 0.342 to 0.351 (+ 2.9 %) | Helped |
| Use sparse updates with update interval increased from 1 to 6 frames (ablation) | ECO · VOT2016 | EAO | Helped |
| Increase the sparse update interval beyond 6 frames (ablation) | ECO · VOT2016 | EAO | Hurt |
| Use a compact sample model instead of standard sample-set management (ablation) | C-COT sample management · VOT2016 | EAO: 0.342 -> 0.351 | Helped |
Closest papers by meaning
- Discriminative Correlation Filter With Channel and Spatial Reliability (CVPR 2017 · similarity 0.89)
- Multi-Cue Correlation Filters for Robust Visual Tracking (CVPR 2018 · similarity 0.89)
- End-To-End Representation Learning for Correlation Filter Based Tracking (CVPR 2017 · similarity 0.88)
- Adaptive Decontamination of the Training Set: A Unified Formulation for Discriminative Visual Tracking (CVPR 2016 · similarity 0.87)
- DiffusionTrack: Diffusion Model for Multi-Object Tracking (AAAI 2024 · similarity 0.87)
- VastTrack: Vast Category Visual Object Tracking (NeurIPS 2024 · similarity 0.86)
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.