VideoCoCa: Video-Text Modeling with Zero-Shot Transfer from Contrastive Captioners
This work addresses the challenge of video-text modeling for AI applications, offering an efficient solution that reuses existing models, though it is incremental as it builds on prior image-text methods.
The authors tackled the problem of building a foundational video-text model efficiently by adapting a pretrained image-text model to video tasks with minimal extra training, achieving state-of-the-art results in zero-shot video classification and text-to-video retrieval.
We explore an efficient approach to establish a foundational video-text model. We present VideoCoCa that maximally reuses a pretrained image-text contrastive captioner (CoCa) model and adapt it to video-text tasks with minimal extra training. While previous works adapt image-text models with various cross-frame fusion modules, we find that the generative attentional pooling and contrastive attentional pooling layers in CoCa are instantly adaptable to flattened frame embeddings, yielding state-of-the-art results on zero-shot video classification and zero-shot text-to-video retrieval. Furthermore, we explore lightweight finetuning on top of VideoCoCa, and achieve strong results on video question-answering and video captioning.