Structuring GUI Elements through Vision Language Models: Towards Action Space Generation
This work addresses a specific bottleneck in GUI understanding for human-computer interaction applications, representing an incremental advancement.
The paper tackles the challenge of generating precise UI element coordinates using multimodal large language models (MLLMs) by introducing an IoU-Augmented Maximum Likelihood (IAML) training paradigm, which improves performance over traditional methods.
Multimodal large language models (MLLMs) have emerged as pivotal tools in enhancing human-computer interaction. In this paper we focus on the application of MLLMs in the field of graphical user interface (GUI) elements structuring, where they assist in processing user instructions based on screen contents. Despite the promise of MLLMs, their performance in precisely generating UI element coordinates, a critical aspect of GUI understanding, is hindered by the nature of next-token prediction training. This challenge arises from the semantic void surrounding numerical UI coordinates in language representation spaces, necessitating a substantial and diverse dataset to bolster visual module capabilities. To address these limitations, we introduce an IoU-Augmented Maximum Likelihood (IAML) training paradigm. Specifically, our approach involves a novel pipeline for IoU-based coordinate sampling to augment the training data, which considers the proximity to ground truth coordinates. This data augmentation strategy is then employed to fine-tune MLLMs under the IAML paradigm, which is designed to mitigate the exposure bias problem inherent in traditional maximum likelihood estimation. Through extensive experiments, we demonstrate the superior performance of our IAML training approach over traditional training paradigms.