Towards Multimodal Understanding via Stable Diffusion as a Task-Aware Feature Extractor
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918087727513600 |
|---|---|
| author | Agarwal, Vatsal Gwilliam, Matthew Kohavi, Gefen Verma, Eshan Ulbricht, Daniel Shrivastava, Abhinav |
| author_facet | Agarwal, Vatsal Gwilliam, Matthew Kohavi, Gefen Verma, Eshan Ulbricht, Daniel Shrivastava, Abhinav |
| contents | Recent advances in multimodal large language models (MLLMs) have enabled image-based question-answering capabilities. However, a key limitation is the use of CLIP as the visual encoder; while it can capture coarse global information, it often can miss fine-grained details that are relevant to the input query. To address these shortcomings, this work studies whether pre-trained text-to-image diffusion models can serve as instruction-aware visual encoders. Through an analysis of their internal representations, we find diffusion features are both rich in semantics and can encode strong image-text alignment. Moreover, we find that we can leverage text conditioning to focus the model on regions relevant to the input question. We then investigate how to align these features with large language models and uncover a leakage phenomenon, where the LLM can inadvertently recover information from the original diffusion prompt. We analyze the causes of this leakage and propose a mitigation strategy. Based on these insights, we explore a simple fusion strategy that utilizes both CLIP and conditional diffusion features. We evaluate our approach on both general VQA and specialized MLLM benchmarks, demonstrating the promise of diffusion models for visual understanding, particularly in vision-centric tasks that require spatial and compositional reasoning. Our project page can be found https://vatsalag99.github.io/mustafar/. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_07106 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Towards Multimodal Understanding via Stable Diffusion as a Task-Aware Feature Extractor Agarwal, Vatsal Gwilliam, Matthew Kohavi, Gefen Verma, Eshan Ulbricht, Daniel Shrivastava, Abhinav Computer Vision and Pattern Recognition Machine Learning Recent advances in multimodal large language models (MLLMs) have enabled image-based question-answering capabilities. However, a key limitation is the use of CLIP as the visual encoder; while it can capture coarse global information, it often can miss fine-grained details that are relevant to the input query. To address these shortcomings, this work studies whether pre-trained text-to-image diffusion models can serve as instruction-aware visual encoders. Through an analysis of their internal representations, we find diffusion features are both rich in semantics and can encode strong image-text alignment. Moreover, we find that we can leverage text conditioning to focus the model on regions relevant to the input question. We then investigate how to align these features with large language models and uncover a leakage phenomenon, where the LLM can inadvertently recover information from the original diffusion prompt. We analyze the causes of this leakage and propose a mitigation strategy. Based on these insights, we explore a simple fusion strategy that utilizes both CLIP and conditional diffusion features. We evaluate our approach on both general VQA and specialized MLLM benchmarks, demonstrating the promise of diffusion models for visual understanding, particularly in vision-centric tasks that require spatial and compositional reasoning. Our project page can be found https://vatsalag99.github.io/mustafar/. |
| title | Towards Multimodal Understanding via Stable Diffusion as a Task-Aware Feature Extractor |
| topic | Computer Vision and Pattern Recognition Machine Learning |
| url | https://arxiv.org/abs/2507.07106 |