Research map / Detection and segmentation / Tracking
Deep Regression Tracking with Shrinkage Loss
Xiankai Lu, Chao Ma, Bingbing Ni, Xiaokang Yang, Ian Reid, Ming-Hsuan Yang
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 |
|---|---|---|---|
| add shrinkage loss instead of square loss (L2) (ablation) | DSLT · OTB-2015 | distance precision: 0.870 -> 0.909 | Improved |
| add shrinkage loss instead of focal loss (L3) (ablation) | DSLT · OTB-2015 | distance precision: 0.887 -> 0.909 | Improved |
| add shrinkage loss instead of online hard negative mining (ablation) | DSLT · OTB-2015 | distance precision: 0.876 -> 0.909 | Improved |
| add shrinkage loss instead of square loss (L2) (ablation) | DSLT · OTB-2015 | overlap success AUC: 0.642 -> 0.660 | Improved |
| add connections between convolutional layers (residual fusion of conv4_3 and conv5_3) (ablation) | DSLT · OTB-2015 | distance precision: 0.879 -> 0.909 | Improved |
| fuse conv3_3 with conv4_3 instead of conv4_3 with conv5_3 (ablation) | DSLT · OTB-2015 | distance precision: 0.909 -> 0.872 | Hurt |
| fuse conv3_3 with conv5_3 instead of conv4_3 with conv5_3 (ablation) | DSLT · OTB-2015 | distance precision: 0.909 -> 0.868 | Hurt |
| shrinkage loss reduces iterations to converge relative to other losses (ablation) | DSLT · OTB-2015 | training iterations to converge | Improved |
| add residual connection of layers to a baseline L2 regression tracker versus CREST | L2 baseline tracker · OTB-2015 | distance precision: 83.8% -> 87.0% | Improved |
Closest papers by meaning
- Deep Object Tracking with Shrinkage Loss (TPAMI 2020 · similarity 0.95)
- Learning Regression and Verification Networks for Robust Long-term Tracking (IJCV 2021 · similarity 0.87)
- Learning Spatial-Aware Regressions for Visual Tracking (CVPR 2018 · similarity 0.86)
- ECO: Efficient Convolution Operators for Tracking (CVPR 2017 · similarity 0.86)
- Probabilistic Regression for Visual Tracking (CVPR 2020 · similarity 0.86)
- VastTrack: Vast Category Visual Object Tracking (NeurIPS 2024 · similarity 0.83)
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.
