Research map / Detection and segmentation / Open-set and few-shot detection
Bridging the Gap between Object and Image-level Representations for Open-Vocabulary Detection
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 |
|---|---|---|---|
| Add RKD to the supervised detector (ablation) | COCO detector · COCO | AP_novel: 1.7 -> 21.2 | Helped |
| Add pseudo-box image-level supervision (ablation) | COCO detector · COCO | AP_novel: 1.7 -> 30.4 | Helped |
| Combine RKD and PIS without weight transfer (ablation) | COCO detector · COCO | AP_novel: 30.4 -> 31.5 | Helped |
| Add weight transfer to RKD + PIS (ablation) | COCO detector · COCO | AP_novel: 31.5 -> 36.6 | Helped |
| Use MViT rather than RPN proposals for L1 distillation (ablation) | COCO detector · COCO | AP_novel: 4.0 -> 16.7 | Helped |
| Add IRM to L1 distillation (ablation) | COCO detector · COCO | AP_novel: 16.7 -> 21.2 | Helped |
| Use class-specific MViT proposals instead of max-size MViT proposals for ILS (ablation) | COCO detector · COCO | AP_novel: 28.9 -> 30.4 | Helped |
| Initialize RKD from the supervised-base detector rather than training from scratch (ablation) | COCO detector · COCO | AP_base: 50.9 -> 54.7 | Helped |
| Remove the MLP skip connection from weight transfer (ablation) | Ours · COCO | AP_novel: 36.6 -> 32.5 | Hurt |
| Remove the MLP skip connection from weight transfer (ablation) | Ours · LVIS | AP_r: 17.1 -> 18.1 | Mixed |
| Use weight transfer rather than naively combining RKD and PIS (ablation) | LVIS detector · LVIS | AP: 22.1 -> 22.8 | Helped |
| Use class-specific MViT proposals rather than max-score RPN proposals for ILS (ablation) | LVIS detector · LVIS | AP_r: 12.8 -> 17.0 | Helped |
Closest papers by meaning
- Exploring Region-Word Alignment in Built-in Detector for Open-Vocabulary Object Detection (CVPR 2024 · similarity 0.91)
- LLMs Meet VLMs: Boost Open Vocabulary Object Detection with Fine-grained Descriptors (ICLR 2024 · similarity 0.91)
- Language-conditioned Detection Transformer (CVPR 2024 · similarity 0.90)
- Weakly Supervised Open-Vocabulary Object Detection (AAAI 2024 · similarity 0.90)
- ProxyDet: Synthesizing Proxy Novel Classes via Classwise Mixup for Open-Vocabulary Object Detection (AAAI 2024 · similarity 0.90)
- Cyclic Contrastive Knowledge Transfer for Open-Vocabulary Object Detection (ICLR 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.