Research map / Detection and segmentation / CNN features
Learning Statistical Texture for Semantic Segmentation
Lanyun Zhu, Deyi Ji, Shiping Zhu, Weihao Gan, Wei Wu, Junjie Yan
Check 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 |
|---|---|---|---|
| Use STLNet instead of SFNet | Cityscapes segmentation with ResNet-101 · Cityscapes test | mIoU (%): 81.8 -> 82.3 | Improved |
| Use STLNet instead of CaC-Net | PASCAL Context segmentation with ResNet-101 · PASCAL Context validation | mIoU (%): 55.4 -> 55.8 | Improved |
| Use STLNet instead of CPNet | ADE20K segmentation with ResNet-101 · ADE20K validation | mIoU (%): 46.27 -> 46.48 | Improved |
| Add shallow layer feature fusion (ablation) | ResNet-101 segmentation model · Cityscapes | mIoU (%): 76.4 -> 77.0 | Improved |
| Add TEM to a model with shallow layer feature fusion (ablation) | ResNet-101 + SLF · Cityscapes | mIoU (%): 77.0 -> 80.3 | Improved |
| Add PTFEM to a model with shallow layer feature fusion (ablation) | ResNet-101 + SLF · Cityscapes | mIoU (%): 77.0 -> 80.0 | Improved |
| Add TEM and PTFEM to a model with shallow layer feature fusion (ablation) | ResNet-101 + SLF · Cityscapes | mIoU (%): 77.0 -> 80.9 | Improved |
| Add TEM and PTFEM to ResNet-101 + ASPP + SLF (ablation) | ResNet-101 + ASPP + SLF · Cityscapes | mIoU (%): 79.3 -> 81.5 | Improved |
| Add the learned graph and feature reassignment in TEM (ablation) | ResNet-101 + ASPP + SLF + TEM · Cityscapes | mIoU (%): 79.9 -> 80.9 | Improved |
| Use multiple PTFEM scales instead of any one tested scale (ablation) | STLNet · Cityscapes | mIoU (%): 81.0, 81.0, 81.1, 81.2 -> 81.5 | Improved |
| Set 1-d QCO to 128 quantization levels (ablation) | STLNet with TEM and PTFEM · Cityscapes | mIoU (%): 81.5 | Improved |
| Increase quantization levels beyond 128 (ablation) | STLNet with TEM and PTFEM · Cityscapes | mIoU (%) | Hurt |
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.
- SFNet — SFNet: Learning Object-Aware Semantic Correspondence (CVPR 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
- Structural and Statistical Texture Knowledge Distillation and Learning for Segmentation (TPAMI 2025 · similarity 0.90)
- Integrating Low-Level Visual Cues for Enhanced Unsupervised Semantic Segmentation (AAAI 2025 · similarity 0.85)
- MaSS13K: A Matting-level Semantic Segmentation Benchmark (CVPR 2025 · similarity 0.84)
- Pixels to Tokens: Token Space Efficient Active Learning for Low-Budget Semantic Segmentation (NeurIPS 2026 · similarity 0.84)
- UNIP: Rethinking Pre-trained Attention Patterns for Infrared Semantic Segmentation (ICLR 2025 · similarity 0.84)
- Learning Hierarchical Image Segmentation For Recognition and By Recognition (ICLR 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.
