Research map / Detection and segmentation / CNN features
Building Detail-Sensitive Semantic Segmentation Networks With Polynomial Pooling
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 |
|---|---|---|---|
| Replace max pooling with adaptive polynomial pooling | VGG segmentation networks · PASCAL VOC 2012 val | mIoU: 53.9* -> 55.1 | Helped |
| Replace max pooling with adaptive polynomial pooling | VGG segmentation networks · Cityscapes val | mIoU: 57.2* -> 58.6 | Helped |
| Replace max pooling with adaptive polynomial pooling | VGG segmentation networks · ADE20K val | mIoU: 25.4* -> 25.9 | Helped |
| Replace the default pooling settings with P-pooling | ResNet segmentation networks · PASCAL VOC 2012 val | mIoU: 60.0 -> 61.1 | Helped |
| Replace the default pooling settings with P-pooling | ResNet segmentation networks · Cityscapes val | mIoU: 54.3 -> 56.3 | Helped |
| Add P-pooling layers | State-of-the-art segmentation models · PASCAL VOC 2012 test | mIoU: 65.1 -> 65.6 | Helped |
| Add P-pooling layers | FCN with InceptionV2-BN · PASCAL VOC 2012 test | mIoU: 60.6 -> 62.3 | Helped |
| Add P-pooling layers | DeepLabv3 with Xception-38 · PASCAL VOC 2012 test | mIoU: 73.8 -> 74.2 | Helped |
| Add P-pooling layers | DeepLab with VGG16-20M · PASCAL VOC 2012 val | mIoU: 67.27 -> 67.95 | Helped |
| Add P-pooling layers | DeepLabv3 with Xception-38 · ADE20K test | mIoU: 33.93 -> 34.70 | Helped |
| Use P-pooling only at pool1 rather than replacing all five max-pooling layers (ablation) | VGG segmentation network · PASCAL VOC 2012 val | mIoU: 53.9 -> 54.9 | Helped |
| Use P-pooling only at pool2 rather than replacing all five max-pooling layers (ablation) | VGG segmentation network · PASCAL VOC 2012 val | mIoU: 53.9 -> 54.7 | Helped |
Closest papers by meaning
- Loss Max-Pooling for Semantic Image Segmentation (CVPR 2017 · similarity 0.91)
- Background-Aware Pooling and Noise-Aware Loss for Weakly-Supervised Semantic Segmentation (CVPR 2021 · similarity 0.90)
- Detail-Preserving Pooling in Deep Networks (CVPR 2018 · similarity 0.89)
- Geometry-Aware Edge Pooling for Graph Neural Networks (NeurIPS 2025 · similarity 0.86)
- Unsupervised Semantic Segmentation Through Depth-Guided Feature Correlation and Sampling (CVPR 2024 · similarity 0.86)
- Towards the Uncharted: Density-Descending Feature Perturbation for Semi-supervised Semantic Segmentation (CVPR 2024 · similarity 0.85)
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.