Robust and Lightweight System for Gait-Based Gender Classification toward Viewing Angle Variations

In computer vision applications, gait-based gender classification is a challenging task as person may walk at various angles with respect to the camera viewpoint. some of viewing angles, person’s limb movement can be occluded from camera, preventing perception features. To solve this problem, study proposes robust and lightweight system for classification. It uses gait energy image (GEI) representing an individual. A discrete cosine transform (DCT) applied on GEI generate feature vector. Further, DCT vector XGBoost classifier performing improve results, parameters are tuned. Finally, results compared other state-of-the-art approaches. The performance proposed evaluated OU-MVLP dataset. experiment show mean CCR (correct rate) 95.33% obtained viewpoints illustrate robustness