Co-Training Transformer with Videos and Images Improves Action Recognitionai.googleblog.com2 pointsRafelMri4 years ago