AI Agent learn human behavior by watching YouTube videos

In October 2017, Google's Machine Perception Research organization announced the release of a large segments designed you-tube video series for AI, named AVA,"atomic visual actions" to help neural networks better train abilities to recognize and interpret human behavior [1]. In a process further described in a corresponding research paper on the project, AVA is derived from long-from video content sourced directly from films and television series featured on YouTube, chosen to reflect a wide variety of human actions as well as a diversity in the race and genders of the humans present in the clips. AVA contains the range of 210,000 actions includes videos of people walking, kicking, hugging, kissing, shaking hands. AVA database contains 57,600 clips of three-second clip short videos, for the tech giant's AI droids to watch and learn from [2] . They believe having AI agents watch thousands of videos of human behavior will help them understand "what humans are doing, what they might do next and what they are trying to achieve".