Prediction model of live weight using deep regression RGB-D images

Predicting live weight helps to monitor animal health, effectively conduct genetic selection and determine optimal slaughter time. On large farms, accurate and expensive industrial scales are used to measure live weight. However, a promising alternative is to estimate live weight by using morphometric measurements of the animal and then applying regression equations linking such measurements to live weight. Manual measurements of animals using a tape measure are time-consuming and stressful for the animals. Therefore, computer vision technology is now increasingly being used for non-contact morphometric measurements. This article proposes a new model for predicting live weight based on image regression using deep learning techniques. It is shown that on real datasets the proposed model achieves weight measurement accuracy with a MAE of 35.5 and MAPE of 8.4 on the test dataset.