Ai-powered Gait Analysis For Early Detection Of Mental Stress Using Mobile Video

Abstract

In Recent Years, Mental Stress Has Become A Growing Concern Due To Its Impact On Both Physical Health And Daily Performance. Traditional Methods Of Detecting Stress Often Rely On Physiological Sensors Or Self-reported Assessments, Which Can Be Intrusive, Expensive, Or Unreliable. In This Study, We Introduce A Novel, Non-invasive Approach That Uses Mobile Video To Analyze Human Gait And Identify Early Signs Of Mental Stress. The System Extracts Body Movement Features From Walking Sequences Using Pose Estimation Techniques And Calculates Specific Gait Parameters Such As Step Width, Hip Motion, And Stride Consistency. A New Composite Vector Is Proposed To Represent Behavioral Changes In Walking Patterns Typically Associated With Mental Stress. These Features Are Then Evaluated Using Machine Learning Algorithms To Classify Stress Levels. The Proposed Method Is Designed To Function In Real-time, Requires Only A Smartphone Camera, And Does Not Rely On Wearable Devices. This Makes It A Practical And Accessible Tool For Mental Health Monitoring, Especially In Everyday Environments Where Privacy And Convenience Are Crucial. The System Also Accounts For Variations In Walking Speed, Camera Angle, And Lighting, Ensuring Consistent Performance Across Diverse Scenarios.

Keywords