LAION 发布 Big Video Dataset：1000 万小时开源视频数据集
LAION drops massive open video dataset with 10 million hours of footage for AI research
LAION 发布 Big Video Dataset（BVD），从 CommonCrawl 的 13 亿视频 URL 中下载了 8000 万个视频，总计 1000 万小时，并提取了 5500 万个带自动生成描述的视频片段和 3 亿张静态图像。
LAION has released the Big Video Dataset (BVD), one of the largest open video datasets for AI research. It draws from 1.3 billion video URLs found in CommonCrawl. The team downloaded 80 million of those videos, totaling 10 million hours, and extracted 55 million clips with auto-generated video and audio descriptions plus 300 million still images.
According to the paper, models trained on BVD outperform comparable models trained on InternVid by up to 2.1 percentage points on common video-to-text benchmarks. The training ties video, audio, and text together, learning which visual content matches which descriptions or sounds.
LAION is releasing the dataset for research only. Legally, the organization can likely point to a 2024 Hamburg Regional Court ruling that allowed it to collect copyrighted content for non-commercial research. LAION asks users to respect the rights of original content creators. The dataset and code are freely available.
LAION
来源：The Decoder：AI News · the-decoder.com