Research map / Detection and segmentation / CNN features
Boosting Semantic Segmentation from the Perspective of Explicit Class Embeddings
Yuhe Liu, Chuanjian Liu, Kai Han, Quan Tang, Zengchang Qin
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 Reconstruction module (without diversity loss) (ablation) | ECENet · ADE20K val | mIoU (ss): 52.5 -> 53.1 | Improved |
| add diversity loss L_div to Feature Reconstruction (ablation) | ECENet · ADE20K val | mIoU (ss): 53.1 -> 53.4 | Improved |
| use gated update instead of plus update for class embeddings (ablation) | Class Updater · ADE20K val | mIoU (ss): 52.7 -> 53.4 | Improved |
| set diversity loss coefficient to 0.2 rather than 0.1 or 0.5 (ablation) | Diversity loss · ADE20K val | mIoU (ss): 53.25 / 53.40 / 52.65 for 0.1 / 0.2 / 0.5 | Mixed |
| apply gated updating to SegViT's randomly initialized class embeddings | SegViT (ViT-Base) · ADE20K val | mIoU (ss): 51.3 -> 51.9 | Improved |
| replace SenFormer with ECENet on same SwinT-Large backbone | SenFormer · ADE20K val | mIoU (ms): 54.2 -> 55.2 | Improved |
| replace SenFormer with ECENet on same SwinT-Large backbone | SenFormer · PASCAL-Context val | mIoU (ms): 64.5 -> 65.9 | Improved |
| replace Mask2Former with ECENet on Cityscapes | Mask2Former · Cityscapes val | mIoU (ms): 84.3 -> 84.5 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- SegViT — SegViT: Semantic Segmentation with Plain Vision Transformers (NeurIPS 2022) · better on 2 of 2 shared results
- SegFormer — SegFormer: A Topic Segmentation Model with Controllable Range of Attention (AAAI 2023) · 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
- CAR: Class-Aware Regularizations for Semantic Segmentation (ECCV 2022 · similarity 0.89)
- Rethinking Semantic Segmentation: A Prototype View (CVPR 2022 · similarity 0.89)
- Replay Without Saving: Prototype Derivation and Distribution Rebalance for Class-Incremental Semantic Segmentation (TPAMI 2025 · similarity 0.89)
- Cs2K: Class-specific and Class-shared Knowledge Guidance for Incremental Semantic Segmentation (ECCV 2024 · similarity 0.89)
- Incrementer: Transformer for Class-Incremental Semantic Segmentation With Knowledge Distillation Focusing on Old Class (CVPR 2023 · similarity 0.88)
- Open-Vocabulary Semantic Segmentation with Image Embedding Balancing (CVPR 2024 · similarity 0.87)
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.
