MouSi: Poly-Visual-Expert Vision-Language Models
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866910311466926080 |
|---|---|
| author | Fan, Xiaoran Ji, Tao Jiang, Changhao Li, Shuo Jin, Senjie Song, Sirui Wang, Junke Hong, Boyang Chen, Lu Zheng, Guodong Zhang, Ming Huang, Caishuang Zheng, Rui Xi, Zhiheng Zhou, Yuhao Dou, Shihan Ye, Junjie Yan, Hang Gui, Tao Zhang, Qi Qiu, Xipeng Huang, Xuanjing Wu, Zuxuan Jiang, Yu-Gang |
| author_facet | Fan, Xiaoran Ji, Tao Jiang, Changhao Li, Shuo Jin, Senjie Song, Sirui Wang, Junke Hong, Boyang Chen, Lu Zheng, Guodong Zhang, Ming Huang, Caishuang Zheng, Rui Xi, Zhiheng Zhou, Yuhao Dou, Shihan Ye, Junjie Yan, Hang Gui, Tao Zhang, Qi Qiu, Xipeng Huang, Xuanjing Wu, Zuxuan Jiang, Yu-Gang |
| contents | Current large vision-language models (VLMs) often encounter challenges such as insufficient capabilities of a single visual component and excessively long visual tokens. These issues can limit the model's effectiveness in accurately interpreting complex visual information and over-lengthy contextual information. Addressing these challenges is crucial for enhancing the performance and applicability of VLMs. This paper proposes the use of ensemble experts technique to synergizes the capabilities of individual visual encoders, including those skilled in image-text matching, OCR, image segmentation, etc. This technique introduces a fusion network to unify the processing of outputs from different visual experts, while bridging the gap between image encoders and pre-trained LLMs. In addition, we explore different positional encoding schemes to alleviate the waste of positional encoding caused by lengthy image feature sequences, effectively addressing the issue of position overflow and length limitations. For instance, in our implementation, this technique significantly reduces the positional occupancy in models like SAM, from a substantial 4096 to a more efficient and manageable 64 or even down to 1. Experimental results demonstrate that VLMs with multiple experts exhibit consistently superior performance over isolated visual encoders and mark a significant performance boost as more experts are integrated. We have open-sourced the training code used in this report. All of these resources can be found on our project website. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2401_17221 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | MouSi: Poly-Visual-Expert Vision-Language Models Fan, Xiaoran Ji, Tao Jiang, Changhao Li, Shuo Jin, Senjie Song, Sirui Wang, Junke Hong, Boyang Chen, Lu Zheng, Guodong Zhang, Ming Huang, Caishuang Zheng, Rui Xi, Zhiheng Zhou, Yuhao Dou, Shihan Ye, Junjie Yan, Hang Gui, Tao Zhang, Qi Qiu, Xipeng Huang, Xuanjing Wu, Zuxuan Jiang, Yu-Gang Computer Vision and Pattern Recognition Artificial Intelligence Computation and Language Machine Learning Current large vision-language models (VLMs) often encounter challenges such as insufficient capabilities of a single visual component and excessively long visual tokens. These issues can limit the model's effectiveness in accurately interpreting complex visual information and over-lengthy contextual information. Addressing these challenges is crucial for enhancing the performance and applicability of VLMs. This paper proposes the use of ensemble experts technique to synergizes the capabilities of individual visual encoders, including those skilled in image-text matching, OCR, image segmentation, etc. This technique introduces a fusion network to unify the processing of outputs from different visual experts, while bridging the gap between image encoders and pre-trained LLMs. In addition, we explore different positional encoding schemes to alleviate the waste of positional encoding caused by lengthy image feature sequences, effectively addressing the issue of position overflow and length limitations. For instance, in our implementation, this technique significantly reduces the positional occupancy in models like SAM, from a substantial 4096 to a more efficient and manageable 64 or even down to 1. Experimental results demonstrate that VLMs with multiple experts exhibit consistently superior performance over isolated visual encoders and mark a significant performance boost as more experts are integrated. We have open-sourced the training code used in this report. All of these resources can be found on our project website. |
| title | MouSi: Poly-Visual-Expert Vision-Language Models |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence Computation and Language Machine Learning |
| url | https://arxiv.org/abs/2401.17221 |