A Vision Based Approach for Real Time Detection of Proximity Hazards around Open Wells: An Application in Open Well Safety Monitoring

Open wells are a major hazard in India’s rural and urban settings, especially for youngsters and unwary adults. The absence of safety barriers surrounding wells frequently causes mishaps, which can occasionally be fatal. This study suggests a brand-new, reasonably priced, vision-based solution that uses a CCTV camera to identify people approaching a well and sound an alarm when they get too close. The system draws boundary boxes and determines the distance between the well and the person using object detection methods to precisely detect and classify both. Accidents may be avoided if an automated alert is set off if the distance between the person and the well drops below a certain level. This method is trained using custom YOLOv8 models for well detection and a pre trained model for person detection. This paper emphasizes the system’s importance in the context of countries like India, where safety infrastructure around wells is often insufficient, and accidental falls are a common hazard.