Rapidata/text-2-video-human-preferences-veo2

Google DeepMind Veo2人类偏好数据集，包含了约45000个人类注释，用于评估视频生成模型在提示对齐、连贯性和视觉偏好方面的表现。数据集通过Rapidata Python API在1小时内收集，包含了视频链接、分数、详细结果、文件名和模型名称等字段。

Google DeepMind Veo2 Human Preferences dataset, containing approximately 45,000 human annotations, used to evaluate the performance of video generation models in terms of prompt alignment, coherence, and visual preference. The dataset was collected within 1 hour using the Rapidata Python API and includes fields such as video links, scores, detailed results, file names, and model names.