Research map / Detection and segmentation / Tracking
Learning Feature Embeddings for Discriminant Model based Tracking
Linyu Zheng, Ming Tang, Yingying Chen, Jinqiao Wang, Hanqing Lu
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 |
|---|---|---|---|
| use original shrinkage loss instead of mean square error loss (ablation) | DCFST-18 · OTB2015 | mean AUC: 0.682 -> 0.698 | Improved |
| use the modified shrinkage loss instead of the original shrinkage loss (ablation) | DCFST-18 · OTB2015 | mean AUC: 0.698 -> 0.709 | Improved |
| increase number of RoI samples N from 361 to 961 (ablation) | DCFST-18 · OTB2015 | mean AUC: 0.692 -> 0.709 | Improved |
| increase number of RoI samples N from 961 to 1369 (ablation) | DCFST-18 · OTB2015 | mean AUC: 0.709 -> 0.707 | No clear effect |
| increase number of RoI samples N from 361 to 961 (ablation) | DCFST-18 · OTB2015 | mean FPS: 46 -> 35 | Hurt |
| learn feature embeddings with closed-form ridge regression solver (DCFST-18) versus ATOM baseline | DCFST-18 · OTB2015 | mean overlap: 0.832 -> 0.872 | Improved |
| DCFST-18 versus ATOM baseline | DCFST-18 · NFS (30 FPS) | mean AUC: 0.584 -> 0.634 | Improved |
| DCFST-18 versus DiMP-18 (closed-form solver vs iterative solver) | DCFST-18 · NFS (30 FPS) | mean AUC: 0.610 -> 0.634 | Improved |
| DCFST-18 versus ATOM baseline | DCFST-18 · GOT10k test | success rate at 0.50: 0.634 -> 0.716 | Improved |
| DCFST-50 versus DiMP-50 | DCFST-50 · GOT10k test | success rate at 0.50: 0.717 -> 0.753 | Mixed |
| DCFST-18 versus ATOM baseline | DCFST-18 · TrackingNet test | success (AUC): 0.703 -> 0.739 | Improved |
| DCFST-50 versus DiMP-50 | DCFST-50 · TrackingNet test | success (AUC): 0.740 -> 0.752 | 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.
- ATOM — ATOM: Accurate Tracking by Overlap Maximization (CVPR 2019) · better on 5 of 5 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
- Learning Discriminative Model Prediction for Tracking (ICCV 2019 · similarity 0.88)
- ECO: Efficient Convolution Operators for Tracking (CVPR 2017 · similarity 0.88)
- Release the Power of Online-Training for Robust Visual Tracking (AAAI 2020 · similarity 0.88)
- CREST: Convolutional Residual Learning for Visual Tracking (ICCV 2017 · similarity 0.88)
- Joint Group Feature Selection and Discriminative Filter Learning for Robust Visual Object Tracking (ICCV 2019 · similarity 0.88)
- High-Performance Discriminative Tracking With Transformers (ICCV 2021 · 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.
