Improving and Evaluating Machine Learning Methods for Forensic Shoeprint Matching

TL;DR

This paper introduces a machine learning pipeline for forensic shoeprint matching that enhances accuracy and generalisability across various challenging scenarios, using edge detection, ICP alignment, and random forest classification.

Contribution

It presents a novel pipeline combining coordinate extraction, alignment, and probabilistic classification, and evaluates its generalisability across diverse shoeprint conditions.

Findings

Models trained on specific scenarios perform well within those scenarios.

Training on diverse scenarios yields comparable accuracy to scenario-specific training.

Models struggle to generalise across different shoeprint conditions.

Abstract

We propose a machine learning pipeline for forensic shoeprint pattern matching that improves on the accuracy and generalisability of existing methods. We extract 2D coordinates from shoeprint scans using edge detection and align the two shoeprints with iterative closest point (ICP). We then extract similarity metrics to quantify how well the two prints match and use these metrics to train a random forest that generates a probabilistic measurement of how likely two prints are to have originated from the same outsole. We assess the generalisability of machine learning methods trained on lab shoeprint scans to more realistic crime scene shoeprint data by evaluating the accuracy of our methods on several shoeprint scenarios: partial prints, prints with varying levels of blurriness, prints with different amounts of wear, and prints from different shoe models. We find that models trained on one type of shoeprint yield extremely high levels of accuracy when tested on shoeprint pairs of the same scenario but fail to generalise to other scenarios. We also discover that models trained on a variety of scenarios predict almost as accurately as models trained on specific scenarios.