Gender recognition from gait using RIT and CIT approaches

Gait analyses have been subject to many researches recent years. Previous studies have shown that, gait is a unique biometric data for each person. Based on this, scientists have realized that it is possible to make gender classification from gait. In this study, the feature vectors were extracted from the RIT´s and CIT´s of the binary silhouette images of human gait scenes. These feature vectors were used in the Support Vector Machine (SVM) and Linear Vector Quantization (LVQ) classifiers for gender recognition. Gait data of 100 persons were divided into k-fold as learning and testing data for cross validation. By using 5 cross folds in trails, in average 95.2% true classification success rate was obtained with LVQ while in average 99.3% true classification success rate was obtained with SVM.