Research map / Detection and segmentation / Tracking
Learning the Model Update for Siamese Trackers
Lichao Zhang, Abel Gonzalez-Garcia, Joost van de Weijer, Martin Danelljan, Fahad Shahbaz Khan
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 |
|---|---|---|---|
| replace linear update with UpdateNet | SiamFC · VOT2018 | EAO: 0.188 -> 0.262 | Improved |
| add skip connection from the initial ground-truth template T_0 to UpdateNet (ablation) | UpdateNet (K=1) · VOT2018 | EAO: 0.205 -> 0.250 | Improved |
| add skip connection from the current template T_i (ablation) | UpdateNet (K=1) · VOT2018 | EAO: 0.205 -> 0.207 | No clear effect |
| add skip connection from the previous accumulated template (ablation) | UpdateNet (K=1) · VOT2018 | EAO: 0.205 -> 0.214 | Improved |
| add multi-stage training (K=3 instead of K=1) (ablation) | UpdateNet · VOT2018 | EAO: 0.250 -> 0.262 | Improved |
| replace offline learned weighted fusion with UpdateNet (ablation) | SiamFC · VOT2018 | EAO: 0.198 -> 0.262 | Improved |
| replace linear update with UpdateNet | DaSiamRPN · VOT2016 | EAO: 0.439 -> 0.481 | Improved |
| replace linear update with UpdateNet | DaSiamRPN · VOT2016 | Robustness (R): 0.262 -> 0.206 | Improved |
| replace linear update with UpdateNet | DaSiamRPN · VOT2016 | Accuracy (A): 0.619 -> 0.610 | Hurt |
| replace linear update with UpdateNet | DaSiamRPN · TrackingNet test | Success (%): 63.8 -> 67.7 | Improved |
| replace linear update with UpdateNet | SiamFC · TrackingNet test | Precision (%): 53.3 -> 53.1 | No clear effect |
| replace linear update with UpdateNet | SiamFC · TrackingNet test | Success (%): 57.1 -> 58.4 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- ATOM — ATOM: Accurate Tracking by Overlap Maximization (CVPR 2019) · better on 0 of 3 shared results
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
- SiamRPN++: Evolution of Siamese Visual Tracking With Very Deep Networks (CVPR 2019 · similarity 0.92)
- Adaptive Siamese Tracking with a Compact Latent Network (TPAMI 2022 · similarity 0.91)
- Deeper and Wider Siamese Networks for Real-Time Visual Tracking (CVPR 2019 · similarity 0.91)
- Discriminative and Robust Online Learning for Siamese Visual Tracking (AAAI 2020 · similarity 0.90)
- GradNet: Gradient-Guided Network for Visual Object Tracking (ICCV 2019 · similarity 0.90)
- Siamese Instance Search for Tracking (CVPR 2016 · similarity 0.89)
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.
