The selection of animals to be marketed is largely completed by their visual assessment, solely relying on the skill level of the animal caretaker. Real-time monitoring of the weight of farm animals would provide important information for not only marketing, but also for the assessment of health and well-being issues. The objective of this study was to develop and evaluate a method based on 3D Convolutional Neural Network to predict weight from point clouds. Intel Real Sense D435 stereo depth camera placed at 2.7 m height was used to capture the 3D videos of a single finishing pig freely walking in a holding pen ranging in weight between 20–120 kg. The animal weight and 3D videos were collected from 249 Landrace × Large White pigs in farm facilities of the FZEA-USP (Faculty of Animal Science and Food Engineering, University of Sao Paulo) between 5 August and 9 November 2021. Point clouds were manually extracted from the recorded 3D video and applied for modeling. A total of 1186 point clouds were used for model training and validating using PointNet framework in Python with a 9:1 split and 112 randomly selected point clouds were reserved for testing. The volume between the body surface points and a constant plane resembling the ground was calculated and correlated with weight to make a comparison with results from the PointNet method. The coefficient of determination (R² = 0.94) was achieved with PointNet regression model on test point clouds compared to the coefficient of determination (R² = 0.76) achieved from the volume of the same animal. The validation RMSE of the model was 6.79 kg with a test RMSE of 6.88 kg. Further, to analyze model performance based on weight range the pigs were divided into three different weight ranges: below 55 kg, between 55 and 90 kg, and above 90 kg. For different weight groups, pigs weighing below 55 kg were best predicted with the model. The results clearly showed that 3D deep learning on point sets has a good potential for accurate weight prediction even with a limited training dataset. Therefore, this study confirms the usability of 3D deep learning on point sets for farm animals’ weight prediction, while a larger data set needs to be used to ensure the most accurate predictions.

Precision livestock farming (PLF) is rapidly advancing, with a growing array of technologies being explored and implemented to improve both productivity and animal welfare. One of the major challenges in this field is the identification of individual animals. Despite numerous efforts having been made to automate this process, there remains a lack of holistic reviews that comprehensively integrate and evaluate these technological developments. Historically, humans have employed various techniques to identify individual animals. This article provides an overview of the evolution of animal identification methods, highlighting significant transitions across various time periods. In prehistoric times, identification relied solely on visual inspection. Today, advanced methods are being utilized, such as radio frequency identification (RFID), computer vision-based systems, biometric recognition, and DNA profiling. Each identification method has its own strengths and limitations. Interestingly, early methods such as visual inspection and drawing can still inspire the development of novel automated systems when combined with modern technologies.