Enhancing Financial VQA in Vision Language Models using Intermediate Structured Representations
This work addresses chart interpretation challenges for automated models in financial visual question answering, but it is incremental as it builds on existing DEPLOT methods with a custom dataset.
This study tackled the problem of accurately extracting information from bar charts by fine-tuning DEPLOT, a modality conversion module, on a custom dataset of 50,000 bar charts, resulting in improved performance on metrics like Relative Mapping Similarity and Relative Number Set Similarity, and showing that providing structured intermediate tables enhances large language model reasoning compared to direct image queries.
Chart interpretation is crucial for visual data analysis, but accurately extracting information from charts poses significant challenges for automated models. This study investigates the fine-tuning of DEPLOT, a modality conversion module that translates the image of a plot or chart to a linearized table, on a custom dataset of 50,000 bar charts. The dataset comprises simple, stacked, and grouped bar charts, targeting the unique structural features of these visualizations. The finetuned DEPLOT model is evaluated against its base version using a test set of 1,000 images and two metrics: Relative Mapping Similarity (RMS), which measures categorical mapping accuracy, and Relative Number Set Similarity (RNSS), which evaluates numerical interpretation accuracy. To further explore the reasoning capabilities of large language models (LLMs), we curate an additional set of 100 bar chart images paired with question answer sets. Our findings demonstrate that providing a structured intermediate table alongside the image significantly enhances LLM reasoning performance compared to direct image queries.