Delving into LLMs’ visual understanding ability using SVG to bridge image and text

Mu Cai, Zeyi Huang, Yuheng Li, Haohan Wang, Yong Jae Lee

Primary Area: representation learning for computer vision, audio, language, and other modalities

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Keywords: Vector representation, Large Language Model

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.

Abstract: Large language models (LLMs) have made significant advancements in natural language understanding. However, through that enormous world model that the LLM has learnt, is it somehow possible for it to understand images as well? This work investigates this question. To enable the LLM to process images, we convert them into a representation given by Scalable Vector Graphics (SVG). To study what the LLM can do with this XML-based textual description of images, we test the LLM on three broad computer vision tasks: visual reasoning, image classification under distribution shift, and generating new images using visual prompting. Even though we do not naturally associate LLMs with any visual understanding capabilities, our results indicate that the LLM can indeed do a pretty decent job in many of these tasks, potentially opening new avenues for research into LLMs ability to understand images.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 4185