VLAB: Enhancing Video Language Pretraining by Feature Adapting and Blending

Large-scale image-text contrastive pre-training models, such as CLIP, have been demonstrated to effectively learn high-quality multimodal representations. However, there is limited research on learning video-text representations for general video multimodal tasks based on these powerful features. Towards this goal, we propose a novel video-text pre-training method dubbed VLAB: <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">V</b>ideo <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">L</b>anguage pre-training by feature <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">A</b>dapting and <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">B</b>lending, which transfers CLIP representations to video pre-training tasks and develops unified video multimodal models for a wide range of video-text tasks. Specifically, VLAB is founded on two key strategies: feature adapting and feature blending. In the former, we introduce a new video adapter module to address CLIP's deficiency in modeling temporal information and extend the model's capability to encompass both contrastive and generative tasks. In the latter, we propose an end-to-end training method that further enhances the model's performance by exploiting the complementarity of image and video features. We validate the effectiveness and versatility of VLAB through extensive experiments on highly competitive video multimodal tasks, including video text retrieval, video captioning, and video question answering. Remarkably, VLAB outperforms competing methods significantly and sets new records in video question answering on MSRVTT, MSVD, and TGIF datasets. It achieves an accuracy of 49.6, 60.9, and 79.0, respectively.