CVPR 2023 Text Guided Video Editing Competition
This addresses the problem of evaluating progress in AI-enabled video editing for researchers and practitioners, but it is incremental as it builds on existing text-to-image models and focuses on benchmarking.
The authors tackled the lack of a standard benchmark for evaluating text-guided video editing models by proposing a new dataset and running a competition at CVPR, resulting in the creation of a publicly available dataset and identification of a winning method.
Humans watch more than a billion hours of video per day. Most of this video was edited manually, which is a tedious process. However, AI-enabled video-generation and video-editing is on the rise. Building on text-to-image models like Stable Diffusion and Imagen, generative AI has improved dramatically on video tasks. But it's hard to evaluate progress in these video tasks because there is no standard benchmark. So, we propose a new dataset for text-guided video editing (TGVE), and we run a competition at CVPR to evaluate models on our TGVE dataset. In this paper we present a retrospective on the competition and describe the winning method. The competition dataset is available at https://sites.google.com/view/loveucvpr23/track4.