A First Look: Towards Explainable TextVQA Models via Visual and Textual Explanations
This addresses the need for explainable AI in applications like e-commerce by providing multimodal explanations that justify model decisions and aid in diagnosing errors, though it is incremental as it builds on prior unimodal explanation methods.
The paper tackles the problem of generating multimodal explanations for TextVQA models by proposing MTXNet, an end-to-end trainable architecture that focuses on text in images, and curating the TextVQA-X dataset, resulting in up to 7% improvement in CIDEr scores and 2% in IoU over unimodal baselines.
Explainable deep learning models are advantageous in many situations. Prior work mostly provide unimodal explanations through post-hoc approaches not part of the original system design. Explanation mechanisms also ignore useful textual information present in images. In this paper, we propose MTXNet, an end-to-end trainable multimodal architecture to generate multimodal explanations, which focuses on the text in the image. We curate a novel dataset TextVQA-X, containing ground truth visual and multi-reference textual explanations that can be leveraged during both training and evaluation. We then quantitatively show that training with multimodal explanations complements model performance and surpasses unimodal baselines by up to 7% in CIDEr scores and 2% in IoU. More importantly, we demonstrate that the multimodal explanations are consistent with human interpretations, help justify the models' decision, and provide useful insights to help diagnose an incorrect prediction. Finally, we describe a real-world e-commerce application for using the generated multimodal explanations.