Research map / Vision-language and multimodal / CLIP and zero-shot
Visual Modality Prompt for Adapting Vision-Language Object Detectors
Heitor R. Medeiros, Atif Belal, Srikanth Muralidharan, Eric Granger, Marco Pedersoli
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 |
|---|---|---|---|
| use ModPrompt encoder-decoder prompt instead of the best static visual prompt (WM) | YOLO-World-S · LLVIP-IR | AP: 50.90 -> 62.87 | Improved |
| use ModPrompt instead of the best static visual prompt | YOLO-World-S · NYUv2-Depth | AP: 6.57 -> 24.93 | Improved |
| use ModPrompt instead of zero-shot | Grounding DINO-T · LLVIP-IR | AP: 56.50 -> 60.10 | Improved |
| use ModPrompt instead of zero-shot | Grounding DINO-T · FLIR-IR | AP: 32.90 -> 32.90 | No clear effect |
| fine-tune the full detector instead of using ModPrompt | YOLO-World-S · COCO retention | AP50: 51.90 -> 00.10 | Hurt |
| replace the MobileNet translator with ResNet in ModPrompt (ablation) | YOLO-World-S · NYUv2-Depth | AP: 23.27 -> 24.93 | Improved |
| replace the MobileNet translator with ResNet in ModPrompt (ablation) | YOLO-World-S · LLVIP-IR | AP: 62.87 -> 61.43 | Hurt |
| increase padding prompt size from 30 to 300 pixels (ablation) | YOLO-World-S · LLVIP-IR | AP: 49.30 -> 19.07 | Hurt |
| increase fixed prompt size from 30 to 300 pixels (ablation) | YOLO-World-S · LLVIP-IR | AP: 37.97 -> 43.53 | No clear effect |
| add MPDR text-embedding residual to ModPrompt (ablation) | YOLO-World-S · LLVIP-IR | AP: 62.87 -> 67.37 | Improved |
| add MPDR text-embedding residual to ModPrompt (ablation) | YOLO-World-S · NYUv2-Depth | AP: 24.93 -> 29.93 | Improved |
| add MPDR to ModPrompt (ablation) | YOLO-World-S · FLIR-IR | AP50: 72.73 (-1.74) | Hurt |
Closest papers by meaning
- On the Test-Time Zero-Shot Generalization of Vision-Language Models: Do We Really Need Prompt Learning? (CVPR 2024 · similarity 0.88)
- LaViP: Language-Grounded Visual Prompting (AAAI 2024 · similarity 0.88)
- Frustratingly Easy Test-Time Adaptation of Vision-Language Models (NeurIPS 2024 · similarity 0.87)
- IPO: Interpretable Prompt Optimization for Vision-Language Models (NeurIPS 2024 · similarity 0.87)
- CoPL: Contextual Prompt Learning for Vision-Language Understanding (AAAI 2024 · similarity 0.87)
- A Closer Look at the Few-Shot Adaptation of Large Vision-Language Models (CVPR 2024 · similarity 0.86)
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.
