Convolution Neural Network (CNN)-Based Live Pig Weight Estimation in Controlled Imaging Platform

This study addresses the need for a more efficient and accurate live pig weight monitoring system in the Indian meat production industry. Conventional methods for measuring pig weights are labor-intensive, prompting the exploration of AI and image processing-based solutions. The research introduces a novel regression-based Convolutional Neural Network (CNN) model trained on a dataset of 1217 images of live pigs, each accompanied by their corresponding weight values. The model demonstrates promising results on the test dataset, with a coefficient of determination (R2) of 0.801, mean absolute error (MAE) of 0.054, and root mean square error (RMSE) of 0.040. Data collection involved a meticulously designed imaging platform to ensure dataset robustness. The proposed model's efficiency is highlighted by its convergence behavior during training and testing, showcasing its ability to accurately predict live pig weights and its potential to revolutionize the Indian meat production industry.