Fake it till you make it: Training Deep Neural Networks for Worker Detection using Synthetic Data

Abstract

The construction industry’s productivity and safety have long been a source of concern, while the broad use of deep neural network (DNN)-based visual AI has transformed other industries. Automation and digitalization powered by DNN provide intriguing answers; yetthe lack of high-quality, diversifieddataprevents the construction sector from leveragingthe benefits. This paper presentsa novel computational framework that enables synthetic data generationfor DNN training to overcome the time-consuming manual data collectionand avoiddata privacy problems. The suggested framework uses graphics engines to create a virtual duplicate of the constructionsite that generates non-real yet realistic visuals. The proposedframework randomizes crucial scene elements such as worker pose, clothes, camera viewpoint, and lighting conditionsto enhancethe variety of the synthetic dataset.The findings of this study presentpromisingpotential of synthetic datain DNN training.

Utilizing 360-Degree Images for Synthetic Data Generation in Construction Scenarios

“…Neuhausen et al, Kim et al, and Tohidifar et al, turned to Blender software, leveraging its capabilities in constructing intricate 3D models of workers (J. Neuhausen et al, 2020;Tohidifar et al, 2022). Similarly, Kim et al extended Blender's utility by creating synthetic images of scaffolds through the integration of point clouds (A. .…”