Research map / Detection and segmentation / Tracking
Beyond Correlation Filters: Learning Continuous Convolution Operators for Visual Tracking
Martin Danelljan, Andreas Robinson, Fahad Shahbaz Khan, Michael Felsberg
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 |
|---|---|---|---|
| fuse layers 0, 1 and 5 instead of layer 1 alone (ablation) | C-COT · OTB-2015 | mean OP: 78.0 -> 82.4 | Improved |
| fuse layers 0, 1 and 5 instead of layer 1 alone (ablation) | C-COT · OTB-2015 | AUC: 65.8 -> 68.2 | Improved |
| add the deep layer 5 to the shallow layer 1 (ablation) | C-COT · OTB-2015 | mean OP: 78.0 -> 81.8 | Improved |
| add the input RGB layer 0 to layers 1 and 5 (ablation) | C-COT · OTB-2015 | mean OP: 81.8 -> 82.4 | Improved |
| add layer 0 to layer 1 (ablation) | C-COT · OTB-2015 | mean OP: 78.0 -> 77.8 | No clear effect |
| use layer 1 instead of layer 5 alone (ablation) | C-COT · OTB-2015 | mean OP: 60.0 -> 78.0 | Improved |
| add intermediate convolutional layers (ablation) | C-COT · OTB-2015 | mean OP | No clear effect |
| use continuous formulation instead of discrete DCF with explicit resampling of layers 0, 1, 5 (ablation) | C-COT · OTB-2015 | AUC: 47.7% -> 68.2% | Improved |
| use continuous formulation instead of discrete DCF resampling layers 0 and 5 to layer 1 resolution (ablation) | C-COT · OTB-2015 | AUC: 60.8% -> 68.2% | Improved |
| resample layers 0 and 5 to layer-1 resolution in discrete DCF instead of resampling all layers to a common lower resolution (ablation) | discrete DCF · OTB-2015 | AUC: 47.7% -> 60.8% | Improved |
| use C-COT instead of DeepSRDCF | C-COT · OTB-2015 | mean OP: 77.3 -> 82.4 | Improved |
| use C-COT instead of DeepSRDCF | C-COT · Temple-Color | mean OP: 65.4 -> 70.4 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Staple — Staple: Complementary Learners for Real-Time Tracking (CVPR 2016) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- ECO: Efficient Convolution Operators for Tracking (CVPR 2017 · similarity 0.92)
- Joint Group Feature Selection and Discriminative Filter Learning for Robust Visual Object Tracking (ICCV 2019 · similarity 0.91)
- Learning Spatial-Temporal Regularized Correlation Filters for Visual Tracking (CVPR 2018 · similarity 0.90)
- Visual Object Tracking with Discriminative Filters and Siamese Networks: A Survey and Outlook (TPAMI 2022 · similarity 0.90)
- Discriminative Correlation Filter Tracker with Channel and Spatial Reliability (IJCV 2018 · similarity 0.90)
- CREST: Convolutional Residual Learning for Visual Tracking (ICCV 2017 · similarity 0.90)
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.
