Synthesizing Reality: Leveraging the Generative AI-Powered Platform Midjourney for Construction Worker Detection

TL;DR

This paper introduces a novel method for generating synthetic construction worker images using Midjourney, significantly improving deep neural network training data diversity and model performance in construction worker detection.

Contribution

It presents a new image synthesis approach using Midjourney to create diverse, realistic datasets for training DNNs in construction worker detection, addressing data scarcity issues.

Findings

Synthetic dataset achieved APs of 0.994 and 0.919.

Model trained on synthetic data performed well on real data.

Demonstrated potential of generative AI to mitigate data scarcity.

Abstract

While recent advancements in deep neural networks (DNNs) have substantially enhanced visual AI's capabilities, the challenge of inadequate data diversity and volume remains, particularly in construction domain. This study presents a novel image synthesis methodology tailored for construction worker detection, leveraging the generative-AI platform Midjourney. The approach entails generating a collection of 12,000 synthetic images by formulating 3000 different prompts, with an emphasis on image realism and diversity. These images, after manual labeling, serve as a dataset for DNN training. Evaluation on a real construction image dataset yielded promising results, with the model attaining average precisions (APs) of 0.937 and 0.642 at intersection-over-union (IoU) thresholds of 0.5 and 0.5 to 0.95, respectively. Notably, the model demonstrated near-perfect performance on the synthetic dataset, achieving APs of 0.994 and 0.919 at the two mentioned thresholds. These findings reveal both the potential and weakness of generative AI in addressing DNN training data scarcity.