Research map / Detection and segmentation / Open-set and few-shot detection
CORA: Adapting CLIP for Open-Vocabulary Detection With Region Prompting and Anchor Pre-Matching
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Apply region prompting to pooled region features | CLIP region classifier · COCO region classification, novel classes | mAP: 58.2 -> 65.1 | Helped |
| Apply region prompting to pooled region features | CLIP RN50x4 region classifier · COCO region classification, novel classes | mAP: 63.9 -> 74.1 | Helped |
| Pool regional features from the image feature map instead of cropping each region as a separate image | CLIP region classifier · COCO region classification, novel classes | mAP: 43.8 -> 58.6 | Helped |
| Add query conditioning on class-name embeddings (ablation) | DETR-style CORA localizer · COCO OVD generalized, novel classes | AP50: 26.0 -> 29.9 | Helped |
| Add per-class post-matching to query conditioning (ablation) | DETR-style CORA localizer · COCO OVD generalized, novel classes | AP50: 29.9 -> 40.9 | Helped |
| Add CLIP-Aligned Labeling (ablation) | CORA with anchor pre-matching · COCO OVD generalized, novel classes | AP50: 40.9 -> 41.7 | Helped |
| Use class dropout during training (ablation) | CORA localizer · COCO OVD | AP50 | Helped |
| Set class dropout probability to 0.2 (ablation) | CORA localizer · COCO OVD | AP50 | Helped |
| Use CORA rather than RegionCLIP for novel-object localization | Open-vocabulary detector · COCO localization, novel classes | AP50: 81.7 -> 83.4 | Helped |
| Train CORA+ with COCO Captions pseudo-box labels | CORA · COCO OVD generalized, novel classes | AP50: 41.7 -> 43.1 | Helped |
Closest papers by meaning
- Simple Image-Level Classification Improves Open-Vocabulary Object Detection (AAAI 2024 · similarity 0.90)
- Exploring Region-Word Alignment in Built-in Detector for Open-Vocabulary Object Detection (CVPR 2024 · similarity 0.90)
- Learning Background Prompts to Discover Implicit Knowledge for Open Vocabulary Object Detection (CVPR 2024 · similarity 0.89)
- Taming Self-Training for Open-Vocabulary Object Detection (CVPR 2024 · similarity 0.89)
- LLMs Meet VLMs: Boost Open Vocabulary Object Detection with Fine-grained Descriptors (ICLR 2024 · similarity 0.89)
- DetCLIPv3: Towards Versatile Generative Open-vocabulary Object Detection (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.