Detection of Manhole and Sewage Overflow Using Computer Vision

Maintenance and management of urban sewage systems are critical for public health and environmental sustainability. Manholes play a crucial role in these systems, providing access points for inspection, maintenance, and environmental protection. Neglected or improperly closed manholes pose significant risks, including accidents, property damage, and environmental pollution. Sewage overflow adds further complications, leading to health hazards and infrastructure damage. This paper presents a deep learning based model for manhole and sewage overflow detection. Utilizing OpenCV for image processing and DarkNet for model integration, proposed system accurately identifies open/closed manholes, and sewage overflows with high precision rates of 97%, 99.87%, and 96.27%, respectively. By automating the maintenance, these solutions contribute to the building of smart cities that enhances safety, reduces operational costs, and safeguards the environment.