Research map / Detection and segmentation / Open-set and few-shot detection
Vocabulary-Free Image Classification and Semantic Segmentation
Alessandro Conti, Enrico Fini, Massimiliano Mancini, Paolo Rota, Yiming Wang, Elisa Ricci
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 prompt ensembling to CaSED | CaSED · Ten classification datasets | Cluster Accuracy (%): 43.1 -> 44.1 | Improved |
| add prompt ensembling to WordNet baseline (ablation) | CLIP + WordNet · Ten classification datasets | Cluster Accuracy (%): 32.6 -> 33.3 | Improved |
| add prompt ensembling to English Words baseline (ablation) | CLIP + English Words · Ten classification datasets | Cluster Accuracy (%): 30.4 -> 29.6 | Hurt |
| add remove operations in candidate filtering (ablation) | CaSED · Ten classification datasets | Cluster Accuracy (%): 27.7 -> 34.1 | Mixed |
| add standardize operations (lowercase, singular) (ablation) | CaSED · Ten classification datasets | Cluster Accuracy (%): 34.1 -> 41.3 | Improved |
| add POS and occurrence filtering (ablation) | CaSED · Ten classification datasets | Cluster Accuracy (%): 41.3 -> 43.1 | Improved |
| retrieve 10 captions instead of 1 (ablation) | CaSED · Ten classification datasets | Cluster Accuracy (%): 30.9 -> 43.1 | Improved |
| retrieve 20 captions instead of 10 (ablation) | CaSED · Ten classification datasets | Cluster Accuracy (%): 43.1 -> 42.9 | No clear effect |
| use SigLIP ViT-L/16 backbone instead of ResNet-50 (ablation) | CaSED · ImageNet (CC12M) | Cluster Accuracy (%): 32.7 -> 47.8 | Improved |
| use ViT-L/14 instead of ResNet-50 (ablation) | CaSED · ImageNet (CC12M) | Cluster Accuracy (%): 32.7 -> 43.5 | Improved |
| use LAION-400M database instead of PMD subset (ablation) | CaSED · ImageNet | Cluster Accuracy (%): 41.5 -> 37.7 | Hurt |
| couple CaSED candidates with SAN instead of English Words | SAN · PascalVOC-20 | Hard Jaccard Index (HJI): 5.8 -> 26.9 | Improved |
Closest papers by meaning
- Training-free Open-Vocabulary Semantic Segmentation via Diverse Prototype Construction and Sub-region Matching (AAAI 2025 · similarity 0.91)
- Emergent Open-Vocabulary Semantic Segmentation from Off-the-shelf Vision-Language Models (CVPR 2024 · similarity 0.91)
- Training-Free Open-Vocabulary Segmentation with Offline Diffusion-Augmented Prototype Generation (CVPR 2024 · similarity 0.90)
- SED: A Simple Encoder-Decoder for Open-Vocabulary Semantic Segmentation (CVPR 2024 · similarity 0.89)
- Image-to-Image Matching via Foundation Models: A New Perspective for Open-Vocabulary Semantic Segmentation (CVPR 2024 · similarity 0.89)
- CLIP as RNN: Segment Countless Visual Concepts without Training Endeavor (CVPR 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.
