Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2512.06663 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908697053102080 |
|---|---|
| author | Qi, Yu Zhang, Yumeng Gong, Chenting Tan, Xiao Zhang, Weiming Zhang, Wei Wang, Jingdong |
| author_facet | Qi, Yu Zhang, Yumeng Gong, Chenting Tan, Xiao Zhang, Weiming Zhang, Wei Wang, Jingdong |
| contents | Large Vision-Language Models (LVLMs) have demonstrated remarkable success in a broad range of vision-language tasks, such as general visual question answering and optical character recognition (OCR). However, their performance on perception-centric tasks -- such as object detection, semantic segmentation, and depth estimation -- remains significantly inferior to that of task-specific expert models. For example, Qwen2.5-VL-7B-Instruct achieves only 19% mAP on COCO2017 val, particularly struggling with dense scenes and small object recall. In this work, we introduce Chain-of-Thought for Detection (CoT4Det), a simple but efficient strategy that reformulates perception tasks into three interpretable steps: classification, counting, and grounding -- each more naturally aligned with the reasoning capabilities of LVLMs. Extensive experiments demonstrate that our method significantly improves perception performance without compromising general vision language capabilities. With a standard Qwen2.5-VL-7B-Instruct, CoT4Det boosts mAP from 19.0% to 33.0% on COCO2017 val and achieves competitive results across a variety of perception benchmarks, outperforming baselines by +2% on RefCOCO series and 19% on Flickr30k entities. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_06663 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | CoT4Det: A Chain-of-Thought Framework for Perception-Oriented Vision-Language Tasks Qi, Yu Zhang, Yumeng Gong, Chenting Tan, Xiao Zhang, Weiming Zhang, Wei Wang, Jingdong Computer Vision and Pattern Recognition Large Vision-Language Models (LVLMs) have demonstrated remarkable success in a broad range of vision-language tasks, such as general visual question answering and optical character recognition (OCR). However, their performance on perception-centric tasks -- such as object detection, semantic segmentation, and depth estimation -- remains significantly inferior to that of task-specific expert models. For example, Qwen2.5-VL-7B-Instruct achieves only 19% mAP on COCO2017 val, particularly struggling with dense scenes and small object recall. In this work, we introduce Chain-of-Thought for Detection (CoT4Det), a simple but efficient strategy that reformulates perception tasks into three interpretable steps: classification, counting, and grounding -- each more naturally aligned with the reasoning capabilities of LVLMs. Extensive experiments demonstrate that our method significantly improves perception performance without compromising general vision language capabilities. With a standard Qwen2.5-VL-7B-Instruct, CoT4Det boosts mAP from 19.0% to 33.0% on COCO2017 val and achieves competitive results across a variety of perception benchmarks, outperforming baselines by +2% on RefCOCO series and 19% on Flickr30k entities. |
| title | CoT4Det: A Chain-of-Thought Framework for Perception-Oriented Vision-Language Tasks |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2512.06663 |