Measurement of pig weight based on computer vision

By analyzing the problems of traditional methods for estimating the weight of pigs, the authors discuss a new method which applies computer vision technology to pig production. The projected image area of pigs when viewed directly from above was computed. The pig weights were estimated by the linear regression of the pig real areas. The results show that a strong relationship exists between pig weight and the projected area of the pig after removing the sections of head and tail in images. The correlation coefficient is 0.94. By comparing with the measured weights, the relative error is less than 2.8%. The experiment indicates that this hands-off method has great significance in scientific management of the pigs, which does not require large labor and material resources, and also avoid the loss in production resulted from stress.