DetPO: In-Context Learning with Multi-Modal LLMs for Few-Shot Object Detection
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914418880675840 |
|---|---|
| author | Gare, Gautam Rajendrakumar Peri, Neehar Popov, Matvei Jain, Shruti Galeotti, John Ramanan, Deva |
| author_facet | Gare, Gautam Rajendrakumar Peri, Neehar Popov, Matvei Jain, Shruti Galeotti, John Ramanan, Deva |
| contents | Multi-Modal LLMs (MLLMs) demonstrate strong visual grounding capabilities on popular object detection benchmarks like OdinW-13 and RefCOCO. However, state-of-the-art models still struggle to generalize to out-of-distribution classes, tasks and imaging modalities not typically found in their pre-training. While in-context prompting is a common strategy to improve performance across diverse tasks, we find that it often yields lower detection accuracy than prompting with class names alone. This suggests that current MLLMs cannot yet effectively leverage few-shot visual examples and rich textual descriptions for object detection. Since frontier MLLMs are typically only accessible via APIs, and state-of-the-art open-weights models are prohibitively expensive to fine-tune on consumer-grade hardware, we instead explore black-box prompt optimization for few-shot object detection. To this end, we propose Detection Prompt Optimization (DetPO), a gradient-free test-time optimization approach that refines text-only prompts by maximizing detection accuracy on few-shot visual training examples while calibrating prediction confidence. Our proposed approach yields consistent improvements across generalist MLLMs on Roboflow20-VL and LVIS, outperforming prior black-box approaches by up to 9.7%. Our code is available at https://github.com/ggare-cmu/DetPO |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_23455 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | DetPO: In-Context Learning with Multi-Modal LLMs for Few-Shot Object Detection Gare, Gautam Rajendrakumar Peri, Neehar Popov, Matvei Jain, Shruti Galeotti, John Ramanan, Deva Computer Vision and Pattern Recognition Multi-Modal LLMs (MLLMs) demonstrate strong visual grounding capabilities on popular object detection benchmarks like OdinW-13 and RefCOCO. However, state-of-the-art models still struggle to generalize to out-of-distribution classes, tasks and imaging modalities not typically found in their pre-training. While in-context prompting is a common strategy to improve performance across diverse tasks, we find that it often yields lower detection accuracy than prompting with class names alone. This suggests that current MLLMs cannot yet effectively leverage few-shot visual examples and rich textual descriptions for object detection. Since frontier MLLMs are typically only accessible via APIs, and state-of-the-art open-weights models are prohibitively expensive to fine-tune on consumer-grade hardware, we instead explore black-box prompt optimization for few-shot object detection. To this end, we propose Detection Prompt Optimization (DetPO), a gradient-free test-time optimization approach that refines text-only prompts by maximizing detection accuracy on few-shot visual training examples while calibrating prediction confidence. Our proposed approach yields consistent improvements across generalist MLLMs on Roboflow20-VL and LVIS, outperforming prior black-box approaches by up to 9.7%. Our code is available at https://github.com/ggare-cmu/DetPO |
| title | DetPO: In-Context Learning with Multi-Modal LLMs for Few-Shot Object Detection |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2603.23455 |