Research map / Detection and segmentation / CNN features
DAVE: A Unified Framework for Fast Vehicle Detection and Annotation
Yi Zhou, Li Liu, Ling Shao, Matt Mellor
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 knowledge guidance from ALN (ablation) | FVPN · UTS | AP: 54.91% -> 57.28% | Improved |
| add knowledge guidance from ALN (ablation) | FVPN · PASCAL VOC2007 car | AP: 58.12% -> 60.05% | Improved |
| add knowledge guidance from ALN (ablation) | FVPN · LISA 2010 | AP: 72.37% -> 73.46% | Improved |
| add bounding-box regression (ablation) | FVPN · UTS | AP: 56.32% -> 57.28% | Improved |
| add bounding-box regression (ablation) | FVPN · PASCAL VOC2007 car | AP: 58.93% -> 60.05% | Improved |
| add bounding-box regression (ablation) | FVPN · LISA 2010 | AP: 71.82% -> 73.46% | Improved |
| add ALN verification after FVPN proposals (ablation) | FVPN · UTS | AP: 57.28% -> 62.85% | Improved |
| add ALN verification after FVPN proposals (ablation) | FVPN · LISA 2010 | AP: 73.46% -> 79.41% | Improved |
| use multi-task instead of single-task learning (ablation) | GoogLeNet ALN · UTS type classification (12 types) | accuracy: 60.37 -> 69.64 | Improved |
| use multi-task instead of single-task learning (ablation) | GoogLeNet ALN · UTS type classification (6 types) | accuracy: 88.32 -> 94.91 | Improved |
| use multi-task instead of single-task learning (ablation) | GoogLeNet ALN · UTS pose estimation | accuracy: 96.94 -> 98.03 | Improved |
| extract features and train SVM on top of single-task features (ablation) | GoogLeNet ALN STL · UTS type classification (6 types) | accuracy: 88.32 -> 90.75 | Mixed |
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.
- FasterRCNN — Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks (TPAMI 2016) · better on 4 of 4 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
- BoxCars: 3D Boxes as CNN Input for Improved Fine-Grained Vehicle Recognition (CVPR 2016 · similarity 0.88)
- FCN-rLSTM: Deep Spatio-Temporal Neural Networks for Vehicle Counting in City Cameras (ICCV 2017 · similarity 0.86)
- Understanding Traffic Density From Large-Scale Web Camera Data (CVPR 2017 · similarity 0.84)
- Learning Deep Neural Networks for Vehicle Re-ID With Visual-Spatio-Temporal Path Proposals (ICCV 2017 · similarity 0.84)
- Deep Relative Distance Learning: Tell the Difference Between Similar Vehicles (CVPR 2016 · similarity 0.84)
- Disentangled Feature Learning Network and a Comprehensive Benchmark for Vehicle Re-Identification (TPAMI 2021 · similarity 0.84)
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.
