Research map / Detection and segmentation / CNN features
Zoom Out-and-In Network with Map Attention Decision for Region Proposal and Object Detection
Hongyang Li, Yu Liu, Wanli Ouyang, Xiaogang Wang
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 |
|---|---|---|---|
| split anchors across three layers instead of all at last feature map (ablation) | zoom-out proposal network · PASCAL VOC region proposals | AR@100: 62.31 -> 65.98 | Improved |
| add zoom-in (deconvolution) path to the zoom-out network (ablation) | ZIP · PASCAL VOC region proposals | AR@100: 65.98 -> 68.51 | Improved |
| place all anchors densely at every layer instead of splitting them (ablation) | zoom-out_sp · PASCAL VOC region proposals | AR@100: 65.98 -> 66.02 | No clear effect |
| stack to 40 layers via zoom-out only instead of zoom-in (ablation) | zoom-out_sp · PASCAL VOC region proposals | AR@100: 68.51 -> 67.21 | Hurt |
| add dynamic alternation of training input scale (ablation) | ZIP · PASCAL VOC region proposals | AR@100: 68.51 -> 72.58 | Improved |
| add equilibrium training strategy (ablation) | ZIP · PASCAL VOC region proposals | AR@100: 68.51 -> 69.85 | Improved |
| add gray-class training strategy (ablation) | ZIP · PASCAL VOC region proposals | AR@100: 68.51 -> 69.77 | Improved |
| add MAD unit on top of all training strategies (ablation) | ZIP + all · PASCAL VOC region proposals | AR@100: 74.22 -> 76.51 | Improved |
| use spatial MAD instead of channel-wise MAD (ablation) | ZIP + all + MAD · PASCAL VOC region proposals | AR@100: 76.51 -> 76.50 | No clear effect |
| replace MAD with SENet excitation recalibration (ablation) | ZIP + all · PASCAL VOC region proposals | AR@100: 76.51 -> 71.87 | Hurt |
| add zoom-in structure to zoom-out detector (ablation) | ZIP detector · PASCAL VOC 2007 test | mAP: 71.3 -> 73.1 | Improved |
| add MAD unit to ZIP detector (ablation) | ZIP baseline · PASCAL VOC 2007 test | mAP: 73.1 -> 76.8 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Faster RCNN — Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks (TPAMI 2016) · better on 2 of 2 shared results
- R-FCN — R-FCN: Object Detection via Region-based Fully Convolutional Networks (NeurIPS 2016) · better on 1 of 1 shared result
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
- Region-Based Convolutional Networks for Accurate Object Detection and Segmentation (TPAMI 2016 · similarity 0.88)
- Feature Selective Networks for Object Detection (CVPR 2018 · similarity 0.87)
- Toward Scale-Invariance and Position-Sensitive Region Proposal Networks (ECCV 2018 · similarity 0.87)
- Learning what and where to attend (ICLR 2019 · similarity 0.87)
- Object Detection Networks on Convolutional Feature Maps (TPAMI 2016 · similarity 0.86)
- Learn to Pay Attention (ICLR 2018 · 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.
