Creating a Forensic Database of Shoeprints from Online Shoe-Tread Photos
Samia Shafique, Bailey Kong, Shu Kong*, and Charless Fowlkes*, 2023
About the Study
Shoe-tread impressions are one of the most common types of evidence left at crime scenes, but research is limited by the lack of databases of footwear prints that cover the large and growing number of distinct shoe models. Moreover, the database is preferred to contain the 3D shape, or depth, of shoe-tread photos so as to allow for extracting shoe prints to match a query (crime-scene) print.
The authors of this study proposed to address this gap by leveraging shoe-tread photos collected by online retailers, but the core challenge was predicting corresponding depth maps. As the images do not have ground-truth 3D shapes allowing for training depth predictors, the authors developed a method, termed ShoeRinsics, that learns to predict depth from fully supervised synthetic data and unsupervised retail image data. In particular, they found that the model yielded significantly better depth predictions than existing methods.
About the Data Set
To validate the method, the authors introduced two validation sets consisting of shoe-tread image and print pairs. The database is freely available to the public.