CW-MobileViT: A Lightweight Deep Learning Model for Cattle Identification in Precision Livestock Farming

With the rapid development of the livestock industry, there is an increasing demand for efficient livestock management and monitoring systems. Simmental cattle, known for their high meat production efficiency, play a crucial role in the global beef supply chain. However, traditional methods for individual identification, such as RFID tags and ear tags, face challenges including high maintenance costs, susceptibility to tag loss, and potential health impacts on livestock. To address these issues, this paper proposes a deep learning-based individual identification model-CW-MobileViT. The model combines the strengths of Convolutional Neural Networks (CNN) and Vision Transformers (ViT), enabling a contactless and automated approach to cattle identification. CW-MobileViT utilizes an enhanced MobileNetV2 backbone, incorporating a Channel Attention mechanism (CA-Attention) to capture both channel and spatial information, thereby improving the representation of individual visual features. Additionally, a weighted cross-entropy loss function is introduced to address the issue of dataset imbalance, ensuring stable training and accurate predictions. Experiments conducted on a self-constructed Simmental cattle dataset demonstrate that the CW-MobileViT model achieves a recognition accuracy of 97.9 %, significantly outperforming existing mainstream models (such as Vision Transformer and ResNet) as well as other lightweight networks (such as MobileNetV2 and RepVGG). The model effectively handles challenges such as individual occlusion and complex background interference in densely populated farming environments. The proposed CW-MobileViT model shows superior performance in cattle identification tasks within intensive farming scenarios, offering a practical solution for precision livestock farming. Future research will focus on real-time optimization of the model and the expansion of the dataset to include more cattle breeds and diversified farming environments.