Research map / Detection and segmentation / CNN features
Learning Implicit Feature Alignment Function for Semantic Segmentation
Hanzhe Hu, Yinbo Chen, Jiarui Xu, Shubhankar Borse, Hong Cai, Fatih Porikli, Xiaolong Wang
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 |
|---|---|---|---|
| replace up-sampling module in FPN with IFA | FPN · Cityscapes val | mIoU: 77.19 -> 78.02 | Improved |
| replace bilinear up-sampling with IFA | DeepLab V3+ · Cityscapes val | mIoU: 76.69 -> 77.57 | Improved |
| replace bilinear up-sampling with IFA | PSPNet · Cityscapes val | mIoU: 73.64 -> 74.42 | Improved |
| replace bilinear up-sampling with IFA | HRNet-W18 · Cityscapes val | mIoU: 77.60 -> 78.10 | Improved |
| replace bilinear up-sampling with IFA | HRNet-W48-OCR · Cityscapes val | mIoU: 85.80 -> 86.10 | Improved |
| add IFA at stride 64 (ablation) | FPN · Cityscapes val (stride 64) | mIoU: 76.52 -> 77.69 | Improved |
| add IFA at stride 128 (ablation) | FPN · Cityscapes val (stride 128) | mIoU: 74.88 -> 76.40 | Improved |
| add IFA at stride 128 (ablation) | DeepLab V3+ · Cityscapes val (stride 128) | mIoU: 70.01 -> 72.18 | Improved |
| add relative coordinates directly in IFA (ablation) | IFA · Cityscapes val | mIoU: +0.2% | Improved |
| use position encoding instead of raw coordinates; learnable frequencies best (ablation) | IFA · Cityscapes val | mIoU | Improved |
| vary position encoding dimension (12 to 72) (ablation) | IFA · Cityscapes val | mIoU | No clear effect |
| add ASPP module to IFA | IFA (ResNet-101) · Cityscapes test | mIoU: 81.2 -> 82.0 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 3 on most shared results.
- SFNet — SFNet: Learning Object-Aware Semantic Correspondence (CVPR 2019) · better on 4 of 5 shared results
- AlignSeg — AlignSeg: Feature-Aligned Segmentation Networks (TPAMI 2021) · better on 0 of 1 shared result
- CARAFE — CARAFE: Content-Aware ReAssembly of FEatures (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
- AlignSeg: Feature-Aligned Segmentation Networks (TPAMI 2021 · similarity 0.87)
- Sfnet: Faster and Accurate Semantic Segmentation Via Semantic Flow (IJCV 2023 · similarity 0.85)
- PCA-Seg: Revisiting Cost Aggregation for Open-Vocabulary Semantic and Part Segmentation (CVPR 2026 · similarity 0.85)
- BAPA-Net: Boundary Adaptation and Prototype Alignment for Cross-Domain Semantic Segmentation (ICCV 2021 · similarity 0.85)
- Implicit Representations for Constrained Image Segmentation (ICML 2024 · similarity 0.84)
- MRFP: Learning Generalizable Semantic Segmentation from Sim-2-Real with Multi-Resolution Feature Perturbation (CVPR 2024 · 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.
