外文摘要：Simple Summary Accurate individual cow identification is key to enhancing farm management efficiency. The mainstream method of identifying cows through physical tags is somewhat invasive. This study introduces a non-invasive approach that utilizes overhead cameras to recognize cows without any physical contact. This method adapts well to changes in cow positions and partial obstructions and does not require system adjustments when new cows are introduced. By focusing on unique features in overhead images, this approach facilitates more convenient and effective monitoring and management of cows, assisting farm operators in achieving efficient daily farm operations.Abstract The automated recognition of individual cows is foundational for implementing intelligent farming. Traditional methods of individual cow recognition from an overhead perspective primarily rely on singular back features and perform poorly for cows with diverse orientation distributions and partial body visibility in the frame. This study proposes an open-set method for individual cow recognition based on spatial feature transformation and metric learning to address these issues. Initially, a spatial transformation deep feature extraction module, ResSTN, which incorporates preprocessing techniques, was designed to effectively address the low recognition rate caused by the diverse orientation distribution of individual cows. Subsequently, by constructing an open-set recognition framework that integrates three attention mechanisms, four loss functions, and four distance metric methods and exploring the impact of each component on recognition performance, this study achieves refined and optimized model configurations. Lastly, introducing moderate cropping and random occlusion strategies during the data-loading phase enhances the model's ability to recognize partially visible individuals. The method proposed in this study achieves a recognition accuracy of 94.58% in open-set scenarios for individual cows in overhead images, with an average accuracy improvement of 2.98 percentage points for cows with diverse orientation distributions, and also demonstrates an improved recognition performance for partially visible and randomly occluded individual cows. This validates the effectiveness of the proposed method in open-set recognition, showing significant potential for application in precision cattle farming management.
外文关键词：open-set recognition;cow individual recognition;overhead perspective;distance metrics;STN
作者：Wang, Buyu;Li, Xia;An, Xiaoping;Duan, Weijun;Wang, Yuan;Wang, Dian;Qi, Jingwei
作者单位：Inner Mongolia Agr Univ;Natl Ctr Technol Innovat Dairy Breeding & Prod Res
期刊名称：ANIMALS
期刊影响因子：0.0
出版年份：2024
出版刊次：14（8）
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