Evaluation of the efficiency of artificial intelligence in estimating the body dimensions of Sistani cows

This research was conducted to investigate the possibility of using visual machine technology in measuring the body dimensions of Sistani cows. For this purpose, the record of body dimensions including length, shoulder height, hip height and chest circumference of 179 heads of livestock in Zahak Sistani cow breeding station, were measured at different time points, using a tape meter. At recording time, digital images were taken using the canon camera from the lateral view of cattle from distance of 2 meters. Digital image processing and feature extraction were performed using Graphical Unit Interference of MATLAB software. The feature of digital images as input and different body dimensions of cows as output of Artificial Neural Network(ANN) were used in the training and modeling phase. The results showed that, out of 22 features extracted from the images of Sistani cows, the 15 effective features, such as Equivalent Diameter, Major Axis Length, Minor Axis Length, Bounding Box, Convex Area, Filled Area, Area, Perimeter, and the number of white pixels of image (NNZ) had a significant correlation with the body dimensions of the Sistani cow(p<0.01). Body dimensions of Sistani cows including body length, shoulder height, hip height, and chest girths were estimated with accuracy of 0.98, 0.97, 0.97and 0.98%, by the ANNs model, respectively. The results of the present research showed that Artificial Intelligence Technology can act as a suitable alternative to biometric evaluation of Sistani cows and save time and relevant costs.