AI-powered gesture recognition for occupational safety improvement in the Sri Lankan construction industry: a conceptual model

Abstract

It is generally accepted that the construction business is one of the most critical industries in which the safety is a paramount need. Even though the traditional safety management measures have been adopted, construction sites still record a high level of occupational accidents. Even though the Artificial Intelligence (AI)-powered gesture recognition has become an option in identifying the unsafe behaviours of workers and facilitating real-time safety interventions, yet it remains a less mature field. As the previous research lacks their focus on AI-based safety monitoring in the construction industry, this paper seeks to conceptualise an AI-powered gesture recognition model to enhance the occupational safety in the Sri Lankan construction sector. A non-systematic literature review was adopted under qualitative phenomenon, drawing on peer-reviewed journals, conference papers, and industry reports from databases including IEEE Xplore, ScienceDirect, Scopus, and Google Scholar, published from 2010 to 2026. The important technological elements, application fields, and implementation strategies of gesture recognition systems in the construction setting were first identified through thematic analysis. Based on these results, an AI-powered gesture recognition model for the Sri Lankan construction sites was conceptualised. The research indicates the possibilities of gesture recognition systems to achieve proactive safety management by real-time identification of dangerous behaviour, better communication, and automatic warnings The study does not involve empirical testing however, it develops a conceptual model based on literature synthesis. Lastly, the paper discusses on the practical issues that are related to the adoption of AI technology in Sri Lankan construction setting and a way forward for future research and practice directions.