Document type : Scientific article published in Applied Animal Behaviour Science
Authors: Qiumei Yang, Deqin Xiao
Preview: Video surveillance technology has been applied in many farms and a large amount of video data has been generated. Extracting valuable information about the pig industry from video data is a challenging problem. Pig behavior recognition is one of the research hotspots. In order to facilitate scholars to systematically understand the research status and development trends of this field, the paper summarizes the research on pig segmentation, pig detection and pig behavior recognition. The applications of traditional image processing methods and deep learning methods are introduced. The recognition strategies for different kinds of behavior, such as postural behavior, locomotion behavior, area-related behavior and interactive behavior, are summarized. Finally, some of the existing challenges in the current research are highlighted and future research work envisioned to address these is proposed.