Pose empowered RGB-flow net

A system comprising:

• data processing hardware;
• and
• memory hardware in communication with the data processing hardware, the memory hardware storing instructions that when executed by the data processing hardware cause the data processing hardware to perform operations comprising:
• receiving video data comprising a series of frames of image data, the video data representative of an actor performing an activity;
• processing the video data to generate:
• a spatial input stream comprising a series of spatial images, each spatial image in the series of spatial images representative of spatial features of the actor performing the activity;
• a temporal input stream representative of motion of the actor performing the activity;
• and
• a pose input stream comprising a series of images, each image in the series of images representative of a pose of the actor performing the activity;
• processing, using at least one neural network, the temporal input stream, the spatial input stream, and the pose input stream;
• and
• classifying, by the at least one neural network, the activity based on the temporal input stream, the spatial input stream, and the pose input stream,
• wherein the at least one neural network comprises a student neural network that processes the spatial input stream and at least one teacher neural network that processes the temporal input stream and the pose input stream.