Safety supervision with Workplace Video SearcH for Infractions (WVSHI)

Workplace safety at construction sites in Singapore remains a critical concern. One of the challenges is ensuring sufficient manpower for continuous safety supervision, which unfortunately can become compromised to reduce costs.

To address these challenges, computer vision technologies are adopted in AI-driven safety systems. These systems include object detection capabilities that ensure workers are equipped with the protective gear, that are needed for safety on site.

With advancements in generative AI technologies, including Large Language Models (LLMs) and Large Multimodal Models (LMMs), WVSHI is working on developing human-level analysis of safety protocols on-site, with the ability to offer real-time recommendations when safety lapses are detected. It serves as a co-pilot for safety inspectors, alleviating their workload, ensuring consistent and thorough inspection through automation. The AI models are specifically designed to understand work process documentation, search large volumes of video footage for novel and untrained-for incidents and generate comprehensive safety reports.

The AI safety inspection system actively monitors worksites for safety infractions, identifying novel patterns or behaviours that existing systems may overlook. Unsafe behaviours can lead to accidents, and this technology aids inspectors in detecting potential issues proactively. Additionally, it can be adapted for other visual supervision tasks, transforming repetitive roles into analytical positions that are can attract younger job seekers.

If you are interested in exploring this technology further or collaborating with us, contact us for more information.