Identification of abnormal gait of pigs based on video analysis

Abstract

Gait Analysis has become a new research field in computer vision. So far, however, contributions to this topic almost exclusively considered the problem of person identification. This study describes an automated algorithm that classifies pig's abnormal gait by utilizing video analysis. The classification algorithm consists of three stages: i) Detection and extraction of the moving pig body and its contour from image sequences; ii) Modeling of pig's forelimb and Extraction of gait information by the joint angles and body points; and iii) Motion analysis and feature extraction for classifying abnormal gait. Eigenvectors were extracted by Fourier analysis on the angle sequence. Then, Support Vector Machine (SVM) classifier is applied to classify normal-abnormal gait. The algorithm was tested on a set of 58 video fragments. The average classification rate was about 90%.