Comparison of synthetic dataset generation methods for medical intervention rooms using medical clothing detection as an example

TL;DR

This paper compares methods for generating synthetic datasets in medical intervention rooms, focusing on clothing detection, to address data privacy issues and improve model accuracy in clinical settings.

Contribution

It evaluates different synthetic data generation techniques, including 3D scanning and domain randomization, demonstrating their effectiveness in closing the reality gap for medical image analysis.

Findings

Structured-Domain-Randomization with Mixed-Reality data achieves 72.0% mAP.

Using 15% of target domain data improves mAP to 80.05%.

Full target domain data yields 83.35% mAP.

Abstract

The availability of real data from areas with high privacy requirements, such as the medical intervention space, is low and the acquisition legally complex. Therefore, this work presents a way to create a synthetic dataset for the medical context, using medical clothing as an example. The goal is to close the reality gap between the synthetic and real data. For this purpose, methods of 3D-scanned clothing and designed clothing are compared in a Domain-Randomization and Structured-Domain-Randomization scenario using an Unreal-Engine plugin or Unity. Additionally a Mixed-Reality dataset in front of a greenscreen and a target domain dataset were used. Our experiments show, that Structured-Domain-Randomization of designed clothing together with Mixed-Reality data provide a baseline achieving 72.0% mAP on a test dataset of the clinical target domain. When additionally using 15% of available target domain train data, the gap towards 100% (660 images) target domain train data could be nearly closed 80.05% mAP (81.95% mAP). Finally we show that when additionally using 100% target domain train data the accuracy could be increased to 83.35% mAP.