Research map / Detection and segmentation / Tracking
Robust Visual Tracking via Hierarchical Convolutional Features
Chao Ma, Jia‐Bin Huang, Xiaokang Yang, 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 soft weight, scale proposals and detection proposals (HCFT* vs HCFT) (ablation) | HCFT · OTB2015 | DP rate (%): 83.7 -> 87.0 | Improved |
| add soft weight, scale proposals and detection proposals (HCFT* vs HCFT) (ablation) | HCFT · OTB2015 | OS rate (%): 65.5 -> 72.0 | Improved |
| add soft weight, scale proposals and detection proposals (HCFT* vs HCFT) (ablation) | HCFT · OTB2013 | DP rate (%): 89.1 -> 92.3 | Improved |
| add soft weight, scale proposals and detection proposals (HCFT* vs HCFT) (ablation) | HCFT · OTB2013 | CLE (pixel): 15.7 -> 9.43 | Improved |
| add target recovery via detection proposals (ablation) | HCFT · OTB2015 occlusion and out-of-view attributes | success rate: +4.0% and +6.1% | Improved |
| use soft weight instead of hard weight for combining layer responses (ablation) | HCFT · OTB2015 | accuracy: +1% | Improved |
| combine response maps hierarchically over conv5-4 and conv4-4 instead of conv5-4 alone (ablation) | HCF (VGG-19) · OTB2015 | DP / success | Improved |
| add conv3-4 to the coarse-to-fine hierarchy (C5-C4-C3 vs C5-C4) (ablation) | HCF (VGG-19) · OTB2015 | DP / success | Improved |
| concatenate conv5, conv4 and conv3 features into one filter instead of hierarchical filters (ablation) | HCF (VGG-19) · OTB2015 | DP / success | Hurt |
| use deeper VGG-19 layers instead of shallow layers C2 and C1 as single features (ablation) | HCF · OTB2015 | DP / success | Improved |
| use ResNet-152 last layer instead of VGG-19 features (ablation) | HCF · OTB2015 | DP / success | Hurt |
| use VGG-19 instead of AlexNet features (ablation) | HCF · OTB2015 | DP / success | Improved |
Closest papers by meaning
- Learning Multi-Domain Convolutional Neural Networks for Visual Tracking (CVPR 2016 · similarity 0.93)
- Target-Aware Deep Tracking (CVPR 2019 · similarity 0.92)
- Robust Object Tracking Based on Temporal and Spatial Deep Networks (ICCV 2017 · similarity 0.91)
- Visual Object Tracking with Discriminative Filters and Siamese Networks: A Survey and Outlook (TPAMI 2022 · similarity 0.90)
- Release the Power of Online-Training for Robust Visual Tracking (AAAI 2020 · similarity 0.90)
- STCT: Sequentially Training Convolutional Networks for Visual Tracking (CVPR 2016 · similarity 0.90)
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.
