Research map / Detection and segmentation / Weakly supervised detection and segmentation
ProNet: Learning to Propose Object-Specific Boxes for Cascaded Neural Networks
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use LSE pooling instead of max pooling (ablation) | ProNet proposal network · PASCAL VOC 2012 val | classification mAP: 83.4 -> 84.8 | Helped |
| Use LSE pooling instead of average pooling (ablation) | ProNet proposal network · PASCAL VOC 2012 val | classification mAP: 81.1 -> 84.8 | Helped |
| Use LSE pooling instead of max pooling (ablation) | ProNet proposal network · PASCAL VOC 2012 val | localization mAP: 72.5 -> 74.8 | Helped |
| Use LSE pooling instead of average pooling (ablation) | ProNet proposal network · PASCAL VOC 2012 val | localization mAP: 62.8 -> 74.8 | Helped |
| Add a verification network after proposal generation (ablation) | ProNet · PASCAL VOC 2012 val | classification mAP: 67.8 -> 69.2 | Helped |
| Add a verification network after proposal generation (ablation) | ProNet · PASCAL VOC 2012 val | localization mAP: 43.5 -> 45.4 | Helped |
| Add a second cascade trained using additional MS COCO data (ablation) | ProNet cascade · PASCAL VOC 2012 val | classification mAP: 88.1 -> 89.0 | Helped |
| Add a second cascade trained using additional MS COCO data (ablation) | ProNet cascade · PASCAL VOC 2012 val | localization mAP: 77.7 -> 78.5 | Helped |
| Use tree-structured category-domain branches instead of a chain (ablation) | ProNet cascade · MS COCO val | classification mAP: 69.2 -> 70.9 | Helped |
| Use tree-structured category-domain branches instead of a chain (ablation) | ProNet cascade · MS COCO val | localization mAP: 45.4 -> 46.4 | Helped |
| Pass more top-scoring proposals to verification, increasing k from 1 to 3 per scale (ablation) | ProNet cascade · PASCAL VOC 2012 val | classification mAP: 87.7 -> 88.1 | Helped |
| Pass more top-scoring proposals to verification, increasing k from 1 to 3 per scale (ablation) | ProNet cascade · PASCAL VOC 2012 val | localization mAP: 76.3 -> 77.1 | Helped |
Closest papers by meaning
- Learning Video Object Segmentation From Static Images (CVPR 2017 · similarity 0.86)
- Weakly Supervised Cascaded Convolutional Networks (CVPR 2017 · similarity 0.86)
- Training Object Class Detectors With Click Supervision (CVPR 2017 · similarity 0.85)
- Feature Pyramid Networks for Object Detection (CVPR 2017 · similarity 0.85)
- Spatially Invariant Unsupervised Object Detection with Convolutional Neural Networks (AAAI 2019 · similarity 0.85)
- Self-Supervised Learning of Intertwined Content and Positional Features for Object Detection (ICML 2025 · 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.