Automated methods for the recognition and analysis of behavior have been increasingly used to investigate individual and collective animal behavior. This includes applications such as tracking individuals in large groups (humans in crowds, animals in herds/swarms/flocks, cars in traffic), mapping observations of an individual to a behavioral space ("what is person X doing"), and learning about the structure of social interactions ("who is doing what with whom"). Recent methods allow processing large amounts of data resulting in surprisingly accurate results. We will look at a variety of supervised and unsupervised approaches with examples from our own research.