Computer Vision
This research aims to develop a computer vision system that detects hazards so that supervisors can be alerted to prevent incidents from occurring. We developed an object detection model built on the TensorFlow platform and we applied image processing techniques using OpenCV library. We have successfully developed a working prototype system based on Faster RCNN to detect workers onsite, detect removal of edge protection and monitor workers working under lifted load.
Workers under load
Workers near the open edge
Computer Vision for Safe Management of COVID-19 Risks
SaRRU is evaluating how our existing computer vision technology can be used to help contractors implement safe distancing and crowd control measures
With the COVID-19 pandemic, safe distancing and prevention of over-crowding had become important workplace safety measures. The existing computer vision-based safety monitoring system developed by NUS SaRRU can be extended to facilitate the monitoring of safety distancing and crowd control measures on any floor with CCTV cameras. The proposed system can identify and alert the supervisor when workers are too near to each other and when there is overcrowding in an area. The event can also be logged for reporting and statistical analysis purposes.