PerceptionGPT: Effectively Fusing Visual Perception into LLM

TL;DR

PerceptionGPT introduces an end-to-end framework that enhances large language models with visual perception capabilities by leveraging token embeddings, lightweight encoders, and decoders, achieving superior performance with less training data and computational resources.

Contribution

The paper presents a novel method that integrates visual perception into LLMs using token embeddings and lightweight modules, reducing training complexity and resource requirements.

Findings

Significant performance improvements over previous methods.

Fewer trainable parameters and GPU hours needed.

Effective handling of multiple visual outputs.

Abstract

The integration of visual inputs with large language models (LLMs) has led to remarkable advancements in multi-modal capabilities, giving rise to visual large language models (VLLMs). However, effectively harnessing VLLMs for intricate visual perception tasks remains a challenge. In this paper, we present a novel end-to-end framework named PerceptionGPT, which efficiently and effectively equips the VLLMs with visual perception abilities by leveraging the representation power of LLMs' token embedding. Our proposed method treats the token embedding of the LLM as the carrier of spatial information, then leverage lightweight visual task encoders and decoders to perform visual perception tasks (e.g., detection, segmentation). Our approach significantly alleviates the training difficulty suffered by previous approaches that formulate the visual outputs as discrete tokens, and enables achieving superior performance with fewer trainable parameters, less training data and shorted training time. Moreover, as only one token embedding is required to decode the visual outputs, the resulting sequence length during inference is significantly reduced. Consequently, our approach enables accurate and flexible representations, seamless integration of visual perception tasks, and efficient handling of a multiple of visual outputs. We validate the effectiveness and efficiency of our approach through extensive experiments. The results demonstrate significant improvements over previous methods with much fewer trainable parameters and GPU hours, which facilitates future research in enabling LLMs with visual perception abilities.