This chapter discuss networks for analysis and recognition of behavior. Behavior refers not only to a combination of actions performed by the subject/initiator, but also to a series of actions taken by the subject/initiator and driven by certain subjective intentions. Behavior is semantically higher-level and at a more abstract level. The recognition of behavior requires not only the detection of actions, but also the analysis of the relationship between the actors and actions, and the establishment of the connection between activities and the environment. Recently, there have been many research works related to behavior recognition that mainly combine neural network and deep learning techniques. From a technical perspective, behavior recognition often requires the integration of various spatiotemporal information. This chapter will introduce three networks for behavior analysis and recognition with the help of different information. The first is a network that combines motion information and contextual information. The second is a network that samples videos in sequential segments to obtain information on motion changes over a long period of time. The third a graph convolutional network for activity and behavior recognition, as well as some improvements made to it.