VideoScore: Building Automatic Metrics to Simulate Fine-grained Human Feedback for Video Generation
This addresses the problem of evaluating video generation models for researchers and developers, providing a proxy for human feedback to track progress and improve models via RLHF, though it is incremental as it builds on existing metric frameworks.
The paper tackles the lack of reliable automatic metrics for video generation by releasing VideoFeedback, a large-scale dataset with human annotations on 37.6K videos, and training VideoScore, which achieves a Spearman correlation of 77.1 with human judges, beating prior metrics by about 50 points.
The recent years have witnessed great advances in video generation. However, the development of automatic video metrics is lagging significantly behind. None of the existing metric is able to provide reliable scores over generated videos. The main barrier is the lack of large-scale human-annotated dataset. In this paper, we release VideoFeedback, the first large-scale dataset containing human-provided multi-aspect score over 37.6K synthesized videos from 11 existing video generative models. We train VideoScore (initialized from Mantis) based on VideoFeedback to enable automatic video quality assessment. Experiments show that the Spearman correlation between VideoScore and humans can reach 77.1 on VideoFeedback-test, beating the prior best metrics by about 50 points. Further result on other held-out EvalCrafter, GenAI-Bench, and VBench show that VideoScore has consistently much higher correlation with human judges than other metrics. Due to these results, we believe VideoScore can serve as a great proxy for human raters to (1) rate different video models to track progress (2) simulate fine-grained human feedback in Reinforcement Learning with Human Feedback (RLHF) to improve current video generation models.