POPEN: Preference-Based Optimization and Ensemble for LVLM-Based Reasoning Segmentation
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917973182119936 |
|---|---|
| author | Zhu, Lanyun Chen, Tianrun Xu, Qianxiong Liu, Xuanyi Ji, Deyi Wu, Haiyang Soh, De Wen Liu, Jun |
| author_facet | Zhu, Lanyun Chen, Tianrun Xu, Qianxiong Liu, Xuanyi Ji, Deyi Wu, Haiyang Soh, De Wen Liu, Jun |
| contents | Existing LVLM-based reasoning segmentation methods often suffer from imprecise segmentation results and hallucinations in their text responses. This paper introduces POPEN, a novel framework designed to address these issues and achieve improved results. POPEN includes a preference-based optimization method to finetune the LVLM, aligning it more closely with human preferences and thereby generating better text responses and segmentation results. Additionally, POPEN introduces a preference-based ensemble method for inference, which integrates multiple outputs from the LVLM using a preference-score-based attention mechanism for refinement. To better adapt to the segmentation task, we incorporate several task-specific designs in our POPEN framework, including a new approach for collecting segmentation preference data with a curriculum learning mechanism, and a novel preference optimization loss to refine the segmentation capability of the LVLM. Experiments demonstrate that our method achieves state-of-the-art performance in reasoning segmentation, exhibiting minimal hallucination in text responses and the highest segmentation accuracy compared to previous advanced methods like LISA and PixelLM. Project page is https://lanyunzhu.site/POPEN/ |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_00640 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | POPEN: Preference-Based Optimization and Ensemble for LVLM-Based Reasoning Segmentation Zhu, Lanyun Chen, Tianrun Xu, Qianxiong Liu, Xuanyi Ji, Deyi Wu, Haiyang Soh, De Wen Liu, Jun Computer Vision and Pattern Recognition Existing LVLM-based reasoning segmentation methods often suffer from imprecise segmentation results and hallucinations in their text responses. This paper introduces POPEN, a novel framework designed to address these issues and achieve improved results. POPEN includes a preference-based optimization method to finetune the LVLM, aligning it more closely with human preferences and thereby generating better text responses and segmentation results. Additionally, POPEN introduces a preference-based ensemble method for inference, which integrates multiple outputs from the LVLM using a preference-score-based attention mechanism for refinement. To better adapt to the segmentation task, we incorporate several task-specific designs in our POPEN framework, including a new approach for collecting segmentation preference data with a curriculum learning mechanism, and a novel preference optimization loss to refine the segmentation capability of the LVLM. Experiments demonstrate that our method achieves state-of-the-art performance in reasoning segmentation, exhibiting minimal hallucination in text responses and the highest segmentation accuracy compared to previous advanced methods like LISA and PixelLM. Project page is https://lanyunzhu.site/POPEN/ |
| title | POPEN: Preference-Based Optimization and Ensemble for LVLM-Based Reasoning Segmentation |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2504.00640 |