Research map / Detection and segmentation / CNN features
RefineNet: Multi-Path Refinement Networks for High-Resolution Semantic Segmentation
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 RefineNet-Res101 instead of DeepLab-v2 (Res101) | Person-Part · Person-Part test | IoU: 64.9 -> 68.6 | Helped |
| Use RefineNet-Res152 instead of Context | RefineNet · NYUDv2 test (40 classes) | IoU: 40.6 -> 46.5 | Helped |
| Use RefineNet-Res152 instead of CSupelec-Res101 | RefineNet · PASCAL VOC 2012 test | mean IoU: 80.2 -> 83.4 | Helped |
| Use RefineNet-Res101 instead of LRR-4x | RefineNet · Cityscapes test | IoU: 71.8 -> 73.6 | Helped |
| Use RefineNet-Res152 instead of DeepLab-v2 (Res101) | RefineNet · PASCAL-Context test | IoU: 45.7 -> 47.3 | Helped |
| Use RefineNet-Res152 instead of Context | RefineNet · SUN-RGBD | IoU: 42.3 -> 45.9 | Helped |
| Use RefineNet-Res152 instead of Cascaded-DilatedNet | RefineNet · ADE20K val | IoU: 34.9 -> 40.7 | Helped |
| Add chained residual pooling (ablation) | ResNet-50 RefineNet · NYUDv2 | IoU: 40.4 -> 42.5 | Helped |
| Add chained residual pooling (ablation) | ResNet-50 RefineNet · Person-Part | IoU: 64.1 -> 65.7 | Helped |
| Add multi-scale evaluation (ablation) | ResNet-50 RefineNet with chained pooling · NYUDv2 | IoU: 42.5 -> 43.8 | Helped |
| Add multi-scale evaluation (ablation) | ResNet-50 RefineNet with chained pooling · Person-Part | IoU: 65.7 -> 67.1 | Helped |
| Increase backbone depth from ResNet-50 to ResNet-101 (ablation) | RefineNet with chained pooling, single-scale · NYUDv2 | IoU: 42.5 -> 43.6 | Helped |
Closest papers by meaning
- Motion-Guided Cascaded Refinement Network for Video Object Segmentation (CVPR 2018 · similarity 0.87)
- Full-Resolution Residual Networks for Semantic Segmentation in Street Scenes (CVPR 2017 · similarity 0.87)
- Referring Image Segmentation via Recurrent Refinement Networks (CVPR 2018 · similarity 0.86)
- Recurrent Scene Parsing With Perspective Understanding in the Loop (CVPR 2018 · similarity 0.86)
- Efficient Piecewise Training of Deep Structured Models for Semantic Segmentation (CVPR 2016 · similarity 0.86)
- ISDNet: Integrating Shallow and Deep Networks for Efficient Ultra-High Resolution Segmentation (CVPR 2022 · 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.