Friday 28 February 2025
For people who struggle with mobility, gait analysis is a crucial tool for diagnosing and treating various health conditions. Traditionally, this process has relied on expensive motion capture systems and specialized personnel, making it inaccessible to many. However, recent advancements in deep learning have enabled the development of markerless gait analysis methods using single-camera videos.
Researchers have been working on creating more accessible solutions that can estimate essential gait parameters from standard video footage. One approach uses a dual- pattern input convolutional transformer network to capture both spatial and temporal dynamics of gait patterns. This system is designed to process skeletal sequence data extracted from original videos, representing each frame as a 2D image.
The proposed model consists of three convolution layers and four fully connected layers. Each convolution layer applies filters to extract features from the input images, while the fully connected layers transform these features into a single scalar output – the predicted gait parameter. The system’s architecture allows it to capture long-term dependencies in spatial-temporal data, making it more effective at extracting motion features from anatomical keypoints.
To evaluate its performance, the model was trained on a publicly available dataset featuring 2,212 video recordings of individuals with cerebral palsy. Results showed that the proposed approach achieved high accuracy in estimating critical metrics such as gait deviation index (GDI), knee flexion angle, and walking cadence. In fact, it reduced mean absolute error by 10.6% and 12.0% for GDI and knee flexion, respectively.
The significance of this work lies in its potential to democratize access to gait analysis. By leveraging standard camera equipment and machine learning algorithms, clinicians can now estimate essential gait parameters without relying on expensive motion capture systems or specialized personnel. This could lead to more widespread adoption of gait analysis in clinical settings, ultimately improving patient outcomes.
One area for future exploration is the application of this technology to larger, real-world datasets. With further refinement and validation, markerless gait analysis could become a valuable tool in the diagnosis and treatment of various health conditions.
Cite this article: “Markerless Gait Analysis: A Breakthrough in Accessibility and Accuracy”, The Science Archive, 2025.
Gait Analysis, Deep Learning, Markerless, Video Footage, Single-Camera, Convolutional Transformer Network, Spatial-Temporal Data, Anatomical Keypoints, Motion Capture Systems, Machine Learning Algorithms
