Gait analysis and identification based on joint information using RGB-depth camera

Several approaches and features have been introduced to analyze gait. Position of skeleton joints in 3-dimensional environment could be used to make gait analysis. The main goal of this paper is to develop an approach that can understand a person's gait cycle being used in various research areas such as social security, and medicine. People's distinctive gait cycle can give information such as neurodegenerative diseases such as Parkinson's disease and Alzheimer's disease. Proposed method focuses on getting information about gait cycle using Microsoft Kinect sensor, training the data with both Random Forest and Multi-layer Perceptron, and then applying it for future references. The obtained accuracy for identification is 81.8% and 87.8% for Random Forest and Multi-layer Perceptron respectively.