Building Future AI News Experiences
Hackathon with The Atlantic & Infactory - Jan 31, Palo Alto
Multi-modal video intelligence
Leverage Licensed Training Data
from Premium News & Sports Sources
Discover and access high-quality, fully licensed video and image datasets from leading media organizations. Train bespoke vision models with confidence.
We partner with major news outlets and sports leagues to deliver training-ready datasets you can use without legal risk.
Clear commercial rights from source. Train and deploy models with confidence.
Pre-labeled with bounding boxes, scene classifications, and detailed metadata
News broadcasts, sports footage, and archival content from trusted publishers
Structured for vision transformers, object detection, and multi-modal training
Curated datasets featuring player tracking, action recognition, and broadcast footage — all cleared for commercial model training.
Browse Datasets →Labeled bounding boxes for people, objects, and scene elements
Temporal annotations for sports plays, gestures, and events
Aligned video, audio transcripts, and text descriptions