2013 - 2015
The automatic understanding of activities performed by humans in videos is of great interest as it allows the monitoring of environments based on the analysis of the interaction between individuals and their behaviors. In this way, new technologies for preventing accidents and identifying suspicious behavior can be developed. Therefore, generating benefits and greater well-being for society. For activities performed by humans to be analyzed automatically, tasks such as detection, recognition, tracking and re-identification of people and the recognition of individual actions must be handled accurately and efficiently. Such tasks comprise the subarea of computer vision called people observation, which deals with the analysis of images and videos containing humans. This project aims to solve problems related to people observation by focusing on robust and discriminative approaches so that the number of inaccurate results is reduced and higher level problems, such as activity recognition, can be solved, thus allowing applications automatic monitoring of environments are developed