AI Insight
LAION-BVD is a large-scale open video dataset comprising 80 million downloaded videos totaling 13.5 million hours, collected from 1.3 billion video URLs sourced through CommonCrawl. The dataset uses content-aware scene detection to extract clips with synthetically generated video and audio captions, designed for multimodal pre-training across video, audio, and image modalities. Models trained on this dataset demonstrate competitive performance on standard video-text and audio-text benchmarks, with performance improving as training and model scale increase.
Why it matters
This dataset significantly expands open access to multimodal video data at an unprecedented scale, providing researchers with a freely available resource for developing and training AI models that can understand and process video, audio, and image content together. The extracted scene-changing frames also offer an alternative source of image-text training data with visual characteristics distinct from typical web image datasets.
Understand the Science
⚠️ Preprint – Noch nicht peer-reviewed
Dieser Artikel wurde noch nicht von unabhängigen Experten begutachtet. Die Ergebnisse sind vorläufig und sollten mit Vorsicht interpretiert werden.
-cross
Abstract: We present LAION-BVD (LAION – Big Video Dataset), a large-scale open video dataset for multimodal learning, which contains 1.3B platform-specific video URLs collected from CommonCrawl. From these, we download 80M videos with a total duration of 13.5 million hours. The dataset is designed for multimodal pre-training across the video, audio, and image modalities. Using content-aware scene detection, we extract clips for which we synthetically generate video and audio captions. Models trained on these data achieve competitive performance on standard video-text and audio-text benchmarks, with consistent improvements as training or model scale increases. Additionally, we explore video frames as an alternative source of image-text data by extracting scene-changing frames. These frames exhibit a visual distribution distinct from standard web image corpora, and models trained on this dataset achieve strong image-text retrieval performance. We release LAION-BVD to the research community. It significantly expands open access to multimodal videos at an unprecedented scale.
Source: LAION-BVD: A 13-Million-Hour Open Video Dataset for Multimodal Pre-training