Research map / Detection and segmentation / Tracking
Unveiling the Power of Deep Tracking
Goutam Bhat, Joakim Johnander, Martin Danelljan, Fahad Shahbaz Khan, Michael Felsberg
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 data augmentation to deep feature model training (ablation) | ResNet deep features · OTB-2015 | AUC: 56.2 -> 61.5 | Improved |
| add blur augmentation (ablation) | deep features · OTB-2015 | AUC: gain of 4% | Improved |
| add flip/blur/other augmentations (except shift) (ablation) | deep features · OTB-2015 | AUC: over 1% improvement | Improved |
| add data augmentation to shallow feature training (ablation) | shallow features · OTB-2015 | AUC: all augmentation types lower tracking performance | Hurt |
| widen the label function (sigma 1/12 -> 1/4) (ablation) | deep features · OTB-2015 | AUC: 56.2 -> 60.5 | Improved |
| widen the label function (ablation) | shallow features · OTB-2015 | AUC: inferior results with higher sigma | Hurt |
| add wider label function on top of data augmentation (ablation) | deep features · OTB-2015 | AUC: 61.5 -> 62.0 | Improved |
| use tailored training plus adaptive fusion instead of ECO | tracker with ResNet-50 · NFS | AUC: 45.7 -> 54.1 | Improved |
| use tailored training plus adaptive fusion instead of ECO | tracker with ResNet-50 · OTB-E | AUC: 74.3 -> 78.0 | Improved |
| use tailored training plus adaptive fusion instead of ECO | tracker with VGG-M · OTB-E | AUC: 74.8 -> 74.2 | No clear effect |
| replace fixed-weight sum fusion with adaptive fusion | deep+shallow tracker · NFS and OTB-E | accuracy (OP 0.75) and robustness (OP 0.5) | Improved |
| replace ECO with the proposed tracker | tracker · VOT2017 | EAO: 0.280 -> 0.378 | Improved |
Closest papers by meaning
- Target-Aware Deep Tracking (CVPR 2019 · similarity 0.89)
- Robust Object Tracking Based on Temporal and Spatial Deep Networks (ICCV 2017 · similarity 0.89)
- Learning Policies for Adaptive Tracking With Deep Feature Cascades (ICCV 2017 · similarity 0.88)
- TrackingNet: A Large-Scale Dataset and Benchmark for Object Tracking in the Wild (ECCV 2018 · similarity 0.88)
- Robust Visual Tracking via Hierarchical Convolutional Features (TPAMI 2018 · similarity 0.88)
- VastTrack: Vast Category Visual Object Tracking (NeurIPS 2024 · similarity 0.86)
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.
