“…To further explore the hotspots and the inter-relationship of these topics, the top ten documents of each of these topics were shortlisted and investigated for more details. Object tracking was primarily conducted with the help of sensors [48,49], computer vision techniques [50][51][52], and other real-time technologies [53]; real-time technologies have been widely adopted for various safety-related monitoring tasks on the job site [48,[54][55][56][57][58][59]; machine learning (or deep learning) algorithms have been commonly utilized to understand and classify accident narratives or safety reports [60][61][62][63] and predict injury severity and outcomes [64][65][66][67]; computer vision techniques have been frequently leveraged for safety equipment monitoring [68][69][70] and unsafe behavior detection [71][72][73]; and structural equation modeling has been generally adopted to understand the relationship between safety and health performance and various other factors like stress [74,75], safety culture and climate [76,77], psycholog- Despite the rapid growth and increasing popularity of the above-mentioned emerging topics, many of the traditional topics have not lost their momentum yet. For example, safety climate has been continuously recognized as a critical factor in safety performance [81,82] and has influenced workers' behaviors through organizational structure [83,84] and supervisors' management …”