Research map / Detection and segmentation / Tracking
Unsupervised Learning of Accurate Siamese 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 temporal cycle training with CPT, region masks, and mask-guided loss reweighting | ULAST unsupervised tracker · VOT2018 | EAO: 0.296 to 0.346 | Helped |
| Add online memory update | ULAST* · TrackingNet | Suc: 0.649 -> 0.654 | Helped |
| Add online memory update | ULAST* · VOT2018 | EAO: 0.347 -> 0.355 | Helped |
| Add a residual connection carrying the initial template into CPT output (ablation) | ULAST with RM and CPT · VOT2018 | EAO: 0.346 -> 0.317 | Hurt |
| Add the region mask to CPT (ablation) | ULAST with CPT · VOT2018 | EAO: 0.310 -> 0.346 | Helped |
| Keep candidate-box coordinate gradients enabled through the region mask (ablation) | ULAST with RM and CPT · VOT2018 | EAO: 0.322 -> 0.346 | Helped |
| Use region mask feature selection instead of Precise RoI-pooling (ablation) | ULAST with CPT · VOT2018 | EAO: 0.301 -> 0.346 | Helped |
| Add mask-guided loss reweighting (ablation) | ULAST cycle training · VOT2018 | EAO: 0.331 to 0.346 | Helped |
| Add mask-guided loss reweighting (ablation) | ULAST legacy-only training · VOT2018 | EAO: 0.284 to 0.296 | Helped |
| Use ImageNet pretraining instead of DenseCL self-supervised pretraining (ablation) | ULAST-on · OTB2015 | Suc: 0.610 -> 0.643 | Helped |
| Use ImageNet pretraining instead of DenseCL self-supervised pretraining (ablation) | ULAST-on · LaSOT | Suc: 0.433 -> 0.442 | Helped |
| Lower the region-mask threshold from 0.5 to 0.0 (ablation) | ULAST · VOT2018 | EAO: 0.327 -> 0.346 | Helped |
Closest papers by meaning
- Unsupervised Deep Tracking (CVPR 2019 · similarity 0.92)
- Progressive Unsupervised Learning for Visual Object Tracking (CVPR 2021 · similarity 0.91)
- SiamRPN++: Evolution of Siamese Visual Tracking With Very Deep Networks (CVPR 2019 · similarity 0.89)
- Learning To Fuse Asymmetric Feature Maps in Siamese Trackers (CVPR 2021 · similarity 0.89)
- Learning a Visual Tracker from a Single Movie without Annotation (AAAI 2019 · similarity 0.89)
- Video Diffusion Models Excel at Tracking Similar-Looking Objects Without Supervision (NeurIPS 2025 · 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.