A Better Use of Audio-Visual Cues: Dense Video Captioning with Bi-modal Transformer

Dense video captioning aims to localize and describe important events in\nuntrimmed videos. Existing methods mainly tackle this task by exploiting only\nvisual features, while completely neglecting the audio track. Only a few prior\nworks have utilized both modalities, yet they show poor results or demonstrate\nthe importance on a dataset with a specific domain. In this paper, we introduce\nBi-modal Transformer which generalizes the Transformer architecture for a\nbi-modal input. We show the effectiveness of the proposed model with audio and\nvisual modalities on the dense video captioning task, yet the module is capable\nof digesting any two modalities in a sequence-to-sequence task. We also show\nthat the pre-trained bi-modal encoder as a part of the bi-modal transformer can\nbe used as a feature extractor for a simple proposal generation module. The\nperformance is demonstrated on a challenging ActivityNet Captions dataset where\nour model achieves outstanding performance. The code is available:\nv-iashin.github.io/bmt\n

Paper

Similar papers

© 2026 NYSGPT2525 LLC