What question did this study set out to answer?

The aim is to develop a framework that enhances forensic shoeprint analysis using deep learning techniques despite limited data.

June 1, 2026Scientific ReportsOpen Access

A unified deep learning application framework for forensic shoeprint analysis under data-limited conditions

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial demonstrates enhanced shoeprint matching in forensic analysis, suggesting broader applications.

The aim is to develop a framework that enhances forensic shoeprint analysis using deep learning techniques despite limited data.
Examined the selection of pretrained models and data augmentation strategies.
Conducted experiments on two shoeprint datasets to evaluate model performance.
Applied few-shot fine-tuning with minimal labeled target-domain data.
Domain-aligned models showed significant improvements in performance stability compared to generic models.
Morphology-preserving generative augmentation led to stable performance improvements.
Few-shot fine-tuning with just 1-3% of labeled data matched the performance of full-data training.

Jang et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark