外文摘要：Artificial intelligence and machine learning have recently been applied to improve agricultural and livestock applications. The precise estimation, recommendations, and performances are the main justifications for using technology. The knowledge that can be gained from animal detection and tracking in videos is useful for monitoring body condition, calving processes, behavior analysis, and individual identification. Accurate animal identification is necessary for monitoring animal welfare, disease prevention, vaccination administration, production supply, and ownership management. In this study, a deep learning-based camera tracking system has been built for businesses where animal welfare is a priority. For this purpose, images of camels in their natural habitat were taken in order to create a dataset. The dataset was split into three categories: training, validation, and testing. It contains 19,081 records from 18 different camels. To identify specific camel faces, this study used deep learning algorithms. The EfficientNetV2B0 algorithm had the highest test accuracy, scoring 98.85% with a validation accuracy of 98.53%. The AI for the camel face recognition task has been validated. The usability of AI on the camel face recognition task was successful in terms of recognition accuracy, and it can be used in place of conventional methods.
外文关键词：deep learning;Smart farming;Camel identification;Face detection;Sustainable livestock farming
作者：Koc, Dilara Gerdan;Koc, Caner;Polat, Havva Eylem;Koc, Atakan
作者单位：Ankara Univ;Adnan Menderes Univ
期刊名称：NEURAL COMPUTING & APPLICATIONS
期刊影响因子：0.0
出版年份：2024
出版刊次：36(6)
原文传递申请：江苏省科技资源（工程技术文献）统筹服务平台