Research map / Architectures and efficiency / CNN design
FaPN: Feature-Aligned Pyramid Network for Dense Image Prediction
Shihua Huang, Zhichao Lu, Ran Cheng, Cheng He
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 feature alignment module (FAM) to FPN (ablation) | FPN (R50) · Cityscapes val | mIoU: 77.4 -> 79.7 | Improved |
| add feature selection module (FSM) on top of FAM (ablation) | FPN + FAM (R50) · Cityscapes val | mIoU: 79.7 -> 80.0 | Improved |
| use FSM instead of SE module after FAM (ablation) | FPN + FAM (R50) · Cityscapes val | mIoU: 78.8 -> 80.0 | Improved |
| use FAM instead of deconvolution upsampling (ablation) | FPN + FSM (R50) · Cityscapes val | mIoU: 76.7 -> 80.0 | Improved |
| apply FAM before feature fusion rather than after (ablation) | FaPN (R50) · Cityscapes val | mIoU: 79.3 -> 80.0 | Improved |
| add an extra 3x3 conv to FPN (ablation) | FPN (R50) · Cityscapes val | mIoU: 77.4 -> 77.5 | No clear effect |
| increase backbone from R50 to R101 | FPN · Cityscapes val | mIoU: 77.4 -> 78.9 | Improved |
| replace FPN with FaPN | Faster R-CNN (R50) · COCO val2017 | AP small: 22.4 -> 24.5 | Improved |
| replace FPN with FaPN | Faster R-CNN (R50) · COCO val2017 | AP large: 49.1 -> 49.1 | No clear effect |
| replace FPN with FaPN | PointRend R50 · Cityscapes val boundary pixels | mean boundary mIoU: 55.9 -> 58.5 | Improved |
| add FAM to real-time FPN (ablation) | FPN (R18) · Cityscapes val | mIoU: 68.6 -> 73.8 | Improved |
| add FSM to real-time FPN + FAM (ablation) | FPN + FAM (R18) · Cityscapes val | mIoU: 73.8 -> 74.2 | Improved |
Closest papers by meaning
- Feature Pyramid Networks for Object Detection (CVPR 2017 · similarity 0.88)
- Fast-iTPN: Integrally Pre-Trained Transformer Pyramid Network With Token Migration (TPAMI 2024 · similarity 0.86)
- A2-FPN: Attention Aggregation Based Feature Pyramid Network for Instance Segmentation (CVPR 2021 · similarity 0.86)
- Construct Effective Geometry Aware Feature Pyramid Network for Multi-Scale Object Detection (AAAI 2022 · similarity 0.86)
- NAS-FPN: Learning Scalable Feature Pyramid Architecture for Object Detection (CVPR 2019 · similarity 0.85)
- GAN-Supervised Dense Visual Alignment (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.
