In this paper, we propose and validate target detection using deep learning and background modeling for sheep detection problem in pasture. First, the video data of sheep in pasture are collected using a mid-infrared camera. Second, the foreground of video is extracted by Gaussian Mixture Model (GMM), and morphological operations are also performed to address the noise and jitter problems of video. Finally, the foreground video is fed into a deep learning network to detect the sheep targets. The experimental results show that the method proposed in this paper can effectively detect sheep in pasture with an average prediction accuracy mAP value of 84.2% and a detection speed of 29.45 fps, which achieves better results and has the practical application value.
Sheep Detection in Grassland Using Deep Learning and Background Modelling
Lect. Notes Electrical Eng.
International Conference on Man-Machine-Environment System Engineering ; 2023 ; Beijing, China October 20, 2023 - October 23, 2023
2023-09-05
6 pages
Article/Chapter (Book)
Electronic Resource
English
Pasture flock , target detection , mid-infrared , Gaussian Mixture Model , deep learning Engineering , Manufacturing, Machines, Tools, Processes , Engineering Economics, Organization, Logistics, Marketing , Artificial Intelligence , Aerospace Technology and Astronautics , Environmental Engineering/Biotechnology
DOAJ | 2024
|Background: Deep Reinforcement Learning
Springer Verlag | 2022
|Sheep profile modelling for automated shearing
TIBKAT | 1981
|