Research map / Detection and segmentation / Tracking
DepthTrack: Unveiling the Power of RGBD Tracking
Song Yan, Jinyu Yang, Jani Käpylä, Feng Zheng, Aleš Leonardis, Joni-Kristian Kämäräinen
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 depth-trained depth branch (DeT) (ablation) | DiMP50 · DepthTrack test (sequence-based) | F-score: 0.405 -> 0.532 | Improved |
| add depth-trained depth branch (DeT) (ablation) | ATOM · DepthTrack test (sequence-based) | F-score: 0.363 -> 0.440 | Improved |
| train depth backbone on depth data instead of reusing RGB-trained DiMP50 on depth (ablation) | DiMP50 depth-only · DepthTrack test (sequence-based) | F-score: 0.206 -> 0.394 | Improved |
| depth-only DeT versus RGB DiMP50 (ablation) | DeT-DiMP50 (D) · DepthTrack test (frame-based) | F-score: 0.395 -> 0.411 | Improved |
| use max pooling instead of convolutional pooling for RGB-D merge (ablation) | DeT-DiMP50 · DepthTrack test (sequence-based) | F-score: 0.496 -> 0.532 | Improved |
| use mean pooling instead of max pooling for RGB-D merge (ablation) | DeT-DiMP50 · DepthTrack test (sequence-based) | F-score: 0.532 -> 0.530 | No clear effect |
| use mean pooling instead of max pooling for RGB-D merge (ablation) | DeT-ATOM · DepthTrack test (sequence-based) | F-score: 0.440 -> 0.457 | Improved |
| any of the three pooling layers versus RGB-only master (ablation) | DeT-ATOM · DepthTrack test (frame-based) | F-score: 0.336 -> 0.411 | Improved |
| add DeT depth branch trained on DepthTrack, tested on CDTB without fine-tuning (ablation) | DiMP50 · CDTB (cross-dataset, sequence-based) | F-score: 0.552 -> 0.642 | Improved |
| add DeT depth branch trained on DepthTrack, tested on CDTB without fine-tuning (ablation) | ATOM · CDTB (cross-dataset, sequence-based) | F-score: 0.542 -> 0.585 | Improved |
| mean pooling instead of max on CDTB (ablation) | DeT-DiMP50 · CDTB (cross-dataset, sequence-based) | F-score: 0.642 -> 0.657 | Improved |
| DeT (short-term, depth-trained) versus long-term RGBD winner | ATCAIS · DepthTrack test (sequence-based) | F-score: 0.476 -> 0.532 | 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 3 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
- Collaborating Vision, Depth, and Thermal Signals for Multi-Modal Tracking: Dataset and Algorithm (NeurIPS 2025 · similarity 0.87)
- Single-Model and Any-Modality for Video Object Tracking (CVPR 2024 · similarity 0.87)
- CSTrack: Enhancing RGB-X Tracking via Compact Spatiotemporal Features (ICML 2025 · similarity 0.87)
- Temporal Adaptive RGBT Tracking with Modality Prompt (AAAI 2024 · similarity 0.87)
- RGBT Tracking via All-layer Multimodal Interactions with Progressive Fusion Mamba (AAAI 2025 · similarity 0.85)
- VastTrack: Vast Category Visual Object Tracking (NeurIPS 2024 · similarity 0.85)
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.
