Research map / Detection and segmentation / Open-set and few-shot detection
CLIPer: Hierarchically Improving Spatial Representation of CLIP for Open-Vocabulary Semantic Segmentation
Lin Sun, Jiale Cao, Jin Xie, Xiaoheng Jiang, Yanwei Pang
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 early-layer fusion to baseline (ablation) | CLIPer baseline · VOC | mIoU: 51.2 -> 61.2 | Improved |
| add fine-grained compensation (Stable Diffusion attention) to baseline (ablation) | CLIPer baseline · VOC | mIoU: 51.2 -> 62.8 | Improved |
| add both early-layer fusion and fine-grained compensation (ablation) | CLIPer baseline · VOC | mIoU: 51.2 -> 69.8 | Improved |
| use early-layer fused attention instead of query-query attention (ablation) | CLIPer last-layer attention · VOC | mIoU: 50.1 -> 58.6 | Improved |
| feed early-layer embeddings to last layer with early-layer attention (ablation) | CLIPer · VOC | mIoU: 58.6 -> 61.2 | Improved |
| add early-layer embeddings to value-value attention (ablation) | value-value attention · VOC | mIoU: 51.2 -> 54.8 | Improved |
| fuse Stable Diffusion heads by matrix multiplication instead of single head (ablation) | fine-grained compensation · VOC | mIoU: 65.3 -> 69.8 | Improved |
| average Stable Diffusion heads instead of multiplying (ablation) | fine-grained compensation · VOC | mIoU: 69.8 -> 64.9 | Hurt |
| add fine-grained compensation (diffusion) to CLIPer* (ablation) | CLIPer* · VOC inference time | time (ms): 47 -> 192 | Hurt |
| CLIPer versus CaR on VOC20 with ViT-L | CLIPer · VOC20 | mIoU: 91.4 -> 90.0 | 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.
- ProxyCLIP — ProxyCLIP: Proxy Attention Improves CLIP for Open-Vocabulary Segmentation (ECCV 2024) · better on 6 of 6 shared results
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
- Unveiling the Knowledge of CLIP for Training-Free Open-Vocabulary Semantic Segmentation (AAAI 2025 · similarity 0.92)
- Enhancing Local Cognition of CLIP for Training-Free Open Vocabulary Semantic Segmentation (TPAMI 2026 · similarity 0.92)
- Explore the Potential of CLIP for Training-Free Open Vocabulary Semantic Segmentation (ECCV 2024 · similarity 0.92)
- Understanding Fine-tuning CLIP for Open-vocabulary Semantic Segmentation in Hyperbolic Space (CVPR 2025 · similarity 0.90)
- CLIPSelf: Vision Transformer Distills Itself for Open-Vocabulary Dense Prediction (ICLR 2024 · similarity 0.89)
- Towards Open-Vocabulary Semantic Segmentation Without Semantic Labels (NeurIPS 2024 · similarity 0.89)
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.
