Multimodal Transformer for Comics Text-Cloze
This work addresses a specific problem in comics analysis for researchers and developers, offering incremental improvements in multimodal understanding.
The paper tackled the Text-cloze task in comics, which involves selecting correct text for a comic panel based on neighboring panels, by introducing a novel Multimodal Large Language Model architecture that achieved a 10% improvement over state-of-the-art models in easy and hard variants, with additional gains from new OCR annotations.
This work explores a closure task in comics, a medium where visual and textual elements are intricately intertwined. Specifically, Text-cloze refers to the task of selecting the correct text to use in a comic panel, given its neighboring panels. Traditional methods based on recurrent neural networks have struggled with this task due to limited OCR accuracy and inherent model limitations. We introduce a novel Multimodal Large Language Model (Multimodal-LLM) architecture, specifically designed for Text-cloze, achieving a 10% improvement over existing state-of-the-art models in both its easy and hard variants. Central to our approach is a Domain-Adapted ResNet-50 based visual encoder, fine-tuned to the comics domain in a self-supervised manner using SimCLR. This encoder delivers comparable results to more complex models with just one-fifth of the parameters. Additionally, we release new OCR annotations for this dataset, enhancing model input quality and resulting in another 1% improvement. Finally, we extend the task to a generative format, establishing new baselines and expanding the research possibilities in the field of comics analysis.