Gait Video

Gait video analysis uses computer vision to extract quantitative information about human movement from video recordings, primarily focusing on improving the assessment and diagnosis of neurological conditions. Current research employs deep learning models, including graph convolutional neural networks and vision-language models, to analyze gait characteristics from both controlled clinical settings and unconstrained home environments, addressing challenges like variable lighting and clothing. This work aims to provide objective, automated, and potentially remote diagnostic tools for conditions like Parkinson's disease and ataxia, improving accessibility and efficiency of clinical assessments. The resulting kinematic data and diagnostic capabilities have significant implications for healthcare, particularly in improving the monitoring and management of movement disorders.