User-Identification-and-Classification-From-Walking-Activity

User personalization and authorization is one of the hottest areas of research as it has widespread applications in the development of security and user personalized applications. Many biometric markers have been used extensively in the past to classify users based upon the features extracted from these markers e.g. one of the most widely used biometric markers for security purposes is the human fingerprint, others worth mentioning are facial and voice features. In this project we propose the use of the human gait recorded through inertial measurement units such as accelerometer to classify users and investigate the gait pattern of different human users to investigate if it is a viable marker for classification and uniqueness due to inherent distinction in the natural walking pattern of human beings. We have used feature extraction algorithms to preprocess and extract features from raw accelerometer data followed by the training of various supervised learning classifiers and evaluating their performance.