作者: Jehan-Antoine Vayssade , Rémy Arquet , Mathieu Bonneau
DOI: 10.1016/J.COMPAG.2019.05.021
关键词: Computer vision 、 Global Positioning System 、 Drone 、 Thresholding 、 Tracking (particle physics) 、 Artificial intelligence 、 Sensitivity (control systems) 、 Activity tracking 、 Process (computing) 、 Common method 、 Computer science
摘要: Abstract Monitoring the position of animals in outdoors can provide useful information ecology and agriculture. A common method is to use active sensors, such as GPS, record their positions at constant intervals time. But using sensors rapidly become expensive when several have be monitored same Another a passive sensor monitor entire flock animals. In this article, we propose process images taken by commercial drone order automate tracking animal activities. We developed that automatically detects goats from tracks activity combination thresholding supervised classification methods. tested our on 571 over 11 days found sensitivity 74% for detection 78.3% detection.