University of Pittsburgh researchers have developed a novel method using computer vision to analyze shoe images and predict traction performance. This technology leverages convolutional neural networks (UNET) to segment shoe outsoles from images and predict the impact of tread features on friction performance. By providing diagnostic feedback on how shoe tread design influences friction, this method could revolutionize the design and evaluation of footwear for various applications.
Description
This invention utilizes a sequence of computer vision steps to predict tread features and mechanical responses on shoe treads, ultimately predicting friction performance. The process involves segmenting the shoe from the background, identifying regions expected to contact the ground, and predicting the impact of these treads on friction performance. The technology can either predict the contact mechanics of the shoe-ground interface or identify tread features associated with good friction performance. This method provides a comprehensive prediction of a shoe's friction performance based on its design characteristics.an application in treating these disorders as well.
Applications
• Footwear design and evaluation
• Sports science and athletic performance
• Safety footwear assessment
• Consumer product testing
Invention Readiness
The method has been developed and tested using convolutional neural networks (UNET) to segment shoe outsoles from images and predict friction performance. The technology is ready for further development and commercialization, with potential applications in footwear design, sports science, and safety assessment.
