AdaMML: Adaptive Multi-Modal Learning for Efficient Video Recognition

Open in new window