The live event will take place on Friday, May 19, 1:00-2:00 PM CT.
Register now to reserve your seat!
Shoe tread impressions are one of the most common types of evidence left at crime scenes. However, the utility of such evidence is limited by the lack of databases of footwear prints that cover the large and growing number of distinct shoe class characteristics. We propose to address this gap by leveraging shoe-tread photos collected by online retailers. The core challenge is to predict the shape and impression pattern from these photos.
We have developed a machine learning-based approach that learns to predict the 3D shape of treads by leveraging a mix of fully supervised synthetic data and unsupervised retail imagery. We quantitatively benchmark the accuracy of our method by comparing predicted pseudo-impressions to a dataset of manually collected test impressions. We also demonstrate initial results in automated matching of crime scene evidence to a database of ~56,000 footwear tread images.
Return to Courses