No More Shoe-Related Back Pain: New St. Petersburg Tech Rewrites Fitting Rules

Engineers from the Saint Petersburg State University of Aerospace Instrumentation (SUAI) have developed a neural network designed to automate the selection of footwear based on individual anatomical parameters. The digital service analyzes a photograph of a user's foot to generate a list of shoe models that match their specific physical characteristics, aiming to reduce errors in choosing athletic and everyday shoes.

Photo: Pravda.Ru by Marina Lebedeva is licensed under Free for commercial use

Deformed foot

The tool targets common fitting difficulties caused by flat feet, hyperpronation, or unusual foot width. According to the developers, wearing improperly fitted footwear can lead to chronic fatigue as well as pain in the joints and back.

Technical Process and Data Scope

To use the algorithm, a user uploads a photo of their foot taken against a light background and provides their gender and personal preferences. The system evaluates three primary metrics: the arch of the foot, the type of pronation, and the width.

The neural network compares these measured variables against a database containing 43 models from six leading global manufacturers. Testing indicates that the service is particularly accurate for users with non-standard anatomy.

During testing, users rated the convenience of the development at 8.5 out of 10. The highest utility was reported by individuals with anatomical deviations who previously spent significant time on manual fittings in stores.

Future Integration

The development team intends to improve existing classifiers and integrate the service into popular marketplace ecosystems to enable direct ordering. This would allow users to purchase footwear based on personalized recommendations without traditional trial-and-error fitting.