google-deepmind/representations4d
Scales self-supervised video models to 22B parameters
View on index · View in 3D Map
Related repos
- Tracks any point in a video
- Dataset to stress-test embedding models
- Large-scale video action recognition dataset
- Learns knowledge from unlabeled videos through generative modeling
- PyTorch models for high-res vision tasks using self-supervised learning
- Survey of LLMs for video understanding
- Trains agents to act from videos
- Labels synchronized videos for AI training
- Open-weights language models from Google DeepMind
- Turn images into cinematic videos with 33 facial expressions and 400+ motion combinations
