Personalized Visual Instruction Tuning
PVIT introduces a framework that curates personalized visual instruction data to cure multimodal models' face blindness, significantly boosting personalized dialogue performance.
Published Oct 9, 2024▲ 70 on Hugging FaceCode ★ 34arXiv ↗
Only vote on papers you've read. Sign in with GitHub to vote.
PVIT makes personalized visual dialogue a credible target for home robots, but its autonomous synthetic pipeline and self-made P-Bench leave it unclear whether it is a true model advance or just face recognition with extra steps.
Abstract
Recent advancements in multimodal large language models (MLLMs) have demonstrated significant progress; however, these models exhibit a notable limitation, which we refer to as "face blindness". Specifically, they can engage in general conversations but fail to conduct personalized dialogues targeting at specific individuals. This deficiency hinders the application of MLLMs in personalized settings, such as tailored visual assistants on mobile devices, or domestic robots that need to recognize members of the family. In this paper, we introduce Personalized Visual Instruction Tuning (PVIT), a novel data curation and training framework designed to enable MLLMs to identify target individuals within an image and engage in personalized and coherent dialogues. Our approach involves the development of a sophisticated pipeline that autonomously generates training data containing personalized conversations. This pipeline leverages the capabilities of various visual experts, image generation models, and (multi-modal) large language models. To evaluate the personalized potential of MLLMs, we present a benchmark called P-Bench, which encompasses various question types with different levels of difficulty. The experiments demonstrate a substantial personalized performance enhancement after fine-tuning with our curated dataset.