Cross-Domain Image Matching with Deep Feature Maps

TL;DR

This paper explores using deep neural network features and a specialized similarity measure to improve cross-domain image matching, with applications in forensic shoeprint identification and other image retrieval tasks.

Contribution

It introduces a multi-channel normalized cross-correlation metric and a discriminative training approach, achieving state-of-the-art results in cross-domain image matching.

Findings

Deep features are effective for cross-domain matching.

Multi-channel normalized cross-correlation improves accuracy.

Discriminative training enhances performance further.

Abstract

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 this 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. Our proposed metric significantly improves performance in matching crime scene shoeprints to laboratory test impressions. We also show its effectiveness in other cross-domain image retrieval problems: matching facade images to segmentation labels and aerial photos to map images. Finally, we introduce a discriminatively trained variant and fine-tune our system through our proposed metric, obtaining state-of-the-art performance.