Cross-Domain Forensic Shoeprint Matching

We investigate the problem of automatically determining what type of shoe left an impression found at a crime scene. This recognition problem is made difficult by the variability in types of crime scene evidence (ranging from traces of dust or oil on hard surfaces to impressions made in soil) and the lack of comprehensive databases of shoe outsole tread patterns. We find that mid-level features extracted by pre-trained convolutional neural nets are surprisingly effective descriptors for these specialized domains. However, the choice of similarity measure for matching exemplars to a query image is essential to good performance. For matching multi-channel deep features, we propose the use of multi-channel normalized cross-correlation and analyze its effectiveness. Finally, we introduce a discriminatively trained variant and fine-tune our system end-to-end, obtaining state-of-the-art performance. Disciplines Forensic Science and Technology Comments This proceeding is published as Kong, Bailey, J. Supancic, Deva Ramanan, and Charless Fowlkes. "Crossdomain forensic shoeprint matching." In British Machine Vision Conference (BMVC), pp. 1-5. 2017. Posted with permission of CSAFE. This conference proceeding is available at Iowa State University Digital Repository: https://lib.dr.iastate.edu/ csafe_conf/3 KONG et al.: CROSS-DOMAIN FORENSIC SHOEPRINT MATCHING 1 Cross-Domain Forensic Shoeprint Matching Bailey Kong1 bhkong@ics.uci.edu James Supancic, III1 jsupanci@uci.edu Deva Ramanan2 deva@andrew.cmu.edu Charless Fowlkes1 fowlkes@ics.uci.edu 1 Dept. of Computer Science University of California, Irvine United States of America 2 Robotics Institute Carnegie Mellon University United States of America