Research map / Detection and segmentation / Open-set and few-shot detection
A Simple Framework for Open-Vocabulary Segmentation and Detection
Hao Zhang, Feng Li, Xueyan Zou, Shilong Liu, Chunyuan Li, Jianwei Yang, Lei Zhang
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 detection data (Objects365) to joint training (ablation) | OpenSeeD Swin-T · COCO panoptic | PQ: 54.0 -> 55.2 | Improved |
| add decoupled foreground and background decoding (ablation) | OpenSeeD-SwinT · ADE20K zero-shot | box AP: 14.9 -> 17.2 | Improved |
| add online mask assistance for matching (ablation) | OpenSeeD-SwinT · ADE20K zero-shot | mask AP: 13.6 -> 14.2 | Improved |
| add offline pseudo masks on Objects365 generated by the large model (ablation) | MaskDINO-R50 pretraining · COCO fine-tune | box AP: 52.0 -> 54.3 | Improved |
| add offline pseudo mask annotations to OpenSeeD-FocalT (ablation) | OpenSeeD-FocalT · ADE20K zero-shot | mask AP: 12.6 -> 14.1 | Improved |
| add offline pseudo annotations to OpenSeeD-FocalT on ADE20K PQ (ablation) | OpenSeeD-FocalT · ADE20K zero-shot | PQ: 18.4 -> 18.2 | No clear effect |
| condition mask decoding on GT concept and box (ablation) | OpenSeeD (T) trained on COCO · ADE20K | mask AP: 8.6 -> 46.4 | Improved |
| use early-stage conditioned masks as direct supervision (ablation) | OpenSeeD online mask assistance · COCO | mask AP: 45.1 -> 42.2 | Hurt |
| OpenSeeD (L) versus X-Decoder (L) on ADE20K panoptic quality | zero-shot segmentation · ADE20K zero-shot | PQ: 21.8 -> 19.7 | Hurt |
| joint segmentation and detection pretraining instead of segmentation-only pretraining | OpenSeeD (T) versus X-Decoder (T) · SegInW | average mask AP: 22.7 -> 33.9 | Improved |
Closest papers by meaning
- A Survey on Open-Vocabulary Detection and Segmentation: Past, Present, and Future (TPAMI 2024 · similarity 0.92)
- Convolutions Die Hard: Open-Vocabulary Segmentation with Single Frozen Convolutional CLIP (NeurIPS 2023 · similarity 0.91)
- Mask-Adapter: The Devil is in the Masks for Open-Vocabulary Segmentation (CVPR 2025 · similarity 0.91)
- Going Denser with Open-Vocabulary Part Segmentation (ICCV 2023 · similarity 0.91)
- Emergent Open-Vocabulary Semantic Segmentation from Off-the-shelf Vision-Language Models (CVPR 2024 · similarity 0.90)
- Training-free Open-Vocabulary Semantic Segmentation via Diverse Prototype Construction and Sub-region Matching (AAAI 2025 · similarity 0.90)
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.
