THUDM/VisionRewardDB-Video
收藏资源简介：
VisionRewardDB-Video数据集是一个综合性的视频评估数据集，旨在为AI生成视频的多维度质量评估提供支持。它包含了21个不同方面的注释，涵盖了文本到视频的一致性、美学质量、运动动力学、物理真实性和技术规格等。数据集经过结构化设计，以支持模型训练和标准化评估，包括主要训练集、带有成对偏好数据的回归集以及用于标准化性能评估的视频偏好测试集。
The VisionRewardDB-Video dataset is a comprehensive collection of video evaluation data designed for multi-dimensional quality assessment of AI-generated videos. It encompasses annotations across 21 diverse aspects, including text-to-video consistency, aesthetic quality, motion dynamics, physical realism, and technical specifications. The dataset is structured to facilitate both model training and standardized evaluation, including a primary training set, a regression set with paired preference data, and a video preference test set for standardized performance evaluation.
