Monocular Visual Pig Weight Estimation Method Based on the EfficientVit-C Model

Abstract

The meat industry is closely related to people’s daily lives and health, and with the growing global population and increasing demand for meat, the development of efficient pig farming technology is particularly important. However, China’s pig industry still faces multiple challenges, such as high labor costs, high biosecurity risks, and low production efficiency. Therefore, there is an urgent need to develop a fast, accurate, and non-invasive method to estimate pig body data to increase production efficiency, enhance biosecurity measures, and improve pig health. This study proposes EfficientVit-C model for image segmentation and cascade several models to estimate the weight of pigs. The EfficientVit-C network uses a cascading group attention module and improves computational efficiency through parameter redistribution and structured pruning. This method uses only one camera for weight estimation, reducing equipment costs and maintenance expenses. The results show that the improved EfficientVit-C model can segment pigs accurately and efficiently the mAP50 curve convergence is 98.2%, the recall is 92.6%, and the precision is 96.5%. The accuracy of pig weight estimation is 100 kg +/− 3.11 kg. On the Jetson Orin NX platform, the average time to complete image segmentation for each 640*480 resolution image was 4.1 ms, and the average time required to complete pig weight estimation was 31 ms. The results show that this method can quickly and accurately estimate the weight of pigs and provide guidance for the subsequent weight evaluation procedures of pigs.