VALUE stands for Video-And-Language Understanding Evaluation, which is a dataset about video and language understanding evaluation. This dataset is a collection of 11 VidL (Video and Language) datasets that can be used for 3 common tasks: text-to-video retrieval; video question answering, and video subtitles. VALUE aims to cover a wide range of video types, video lengths, data volumes, and task difficulty levels. VALUE not only focuses on single-channel video visual information, but also promotes models that utilize video frames and their associated subtitle information and models that share knowledge across multiple tasks.

The datasets used for the VALUE dataset benchmark are:TVQA , TVR , TVC , How2R , How2QA , VIOLIN , VLEP , YouCook2(YC2C, YC2R), VATEX