English

TxT: Crossmodal End-to-End Learning with Transformers

Computer Vision and Pattern Recognition 2021-09-10 v1 Computation and Language

Abstract

Reasoning over multiple modalities, e.g. in Visual Question Answering (VQA), requires an alignment of semantic concepts across domains. Despite the widespread success of end-to-end learning, today's multimodal pipelines by and large leverage pre-extracted, fixed features from object detectors, typically Faster R-CNN, as representations of the visual world. The obvious downside is that the visual representation is not specifically tuned to the multimodal task at hand. At the same time, while transformer-based object detectors have gained popularity, they have not been employed in today's multimodal pipelines. We address both shortcomings with TxT, a transformer-based crossmodal pipeline that enables fine-tuning both language and visual components on the downstream task in a fully end-to-end manner. We overcome existing limitations of transformer-based detectors for multimodal reasoning regarding the integration of global context and their scalability. Our transformer-based multimodal model achieves considerable gains from end-to-end learning for multimodal question answering.

Keywords

Cite

@article{arxiv.2109.04422,
  title  = {TxT: Crossmodal End-to-End Learning with Transformers},
  author = {Jan-Martin O. Steitz and Jonas Pfeiffer and Iryna Gurevych and Stefan Roth},
  journal= {arXiv preprint arXiv:2109.04422},
  year   = {2021}
}

Comments

To appear at the 43rd DAGM German Conference on Pattern Recognition (GCPR) 2021

R2 v1 2026-06-24T05:50:05.925Z