Research map / Detection and segmentation / CNN features
Disentangled Non-local Neural Networks
Minghao Yin, Zhuliang Yao, Yue Cao, Xiu Li, Zheng Zhang, Stephen Lin, Han Hu
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 standard non-local block | dilated FCN R101 · Cityscapes val | mIoU: 75.8 -> 78.5 | Improved |
| replace NL block by DNL block (both modifications) (ablation) | NL · Cityscapes val | mIoU: 78.5 -> 80.5 | Improved |
| change multiplication of terms to addition only (ablation) | NL · Cityscapes val | mIoU: 78.5 -> 79.2 | Improved |
| use non-shared key transform for unary term only (ablation) | NL · Cityscapes val | mIoU: 78.5 -> 79.0 | Improved |
| use pairwise term alone (ablation) | dilated FCN R101 · Cityscapes val | mIoU: 75.8 -> 77.5 | Improved |
| use unary term alone instead of both terms in NL (ablation) | NL · Cityscapes val | mIoU: 78.5 -> 79.3 | Improved |
| replace NL with DNL | ResNet-101 · Cityscapes test | mIoU: 80.8 -> 82.0 | Improved |
| replace NL with DNL | HRNetV2-W48 · Cityscapes test | mIoU: 82.5 -> 83.0 | Improved |
| replace NL with DNL | ResNet-101 · ADE20K val | mIoU: 44.67 -> 45.97 | Improved |
| replace NL with DNL | ResNet-101 · PASCAL-Context test | mIoU: 50.6 -> 54.8 | Improved |
| replace NL with DNL | Mask R-CNN R50-FPN · COCO 2017 val | AP bbox: 39.6 -> 40.3 | Improved |
| replace NL with DNL | Slow-only R50 · Kinetics val | top-1 accuracy: 75.95 -> 76.31 | Improved |
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.
- ACFNet — ACFNet: Attentional Class Feature Network for Semantic Segmentation (ICCV 2019) · 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
- Asymmetric Non-Local Neural Networks for Semantic Segmentation (ICCV 2019 · similarity 0.86)
- Non-Local Neural Networks (CVPR 2018 · similarity 0.85)
- Unifying Nonlocal Blocks for Neural Networks (ICCV 2021 · similarity 0.85)
- Poly-NL: Linear Complexity Non-Local Layers With 3rd Order Polynomials (ICCV 2021 · similarity 0.85)
- Non-Salient Region Object Mining for Weakly Supervised Semantic Segmentation (CVPR 2021 · similarity 0.84)
- Compact Generalized Non-local Network (NeurIPS 2018 · 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.
