Abstract
Water accumulation in tunnel threatens the safety of driving and the tunnel itself. In order to detect the tunnel waterlogging in time, a method based on computer vision is proposed. This method utilizes Laplace transform for image preprocessing to remove fuzzy image, uses the network MobileNetV2 to build a tunnel waterlogging recognition model, and then smooths the prediction results. Based on this method, a tunnel waterlogging recognition and early warning platform is developed which applied to a city video surveillance system successfully. The results show that: the tunnel waterlogging recognition method based on computer vision has high accuracy, tiny computation and low cost. The existing video monitoring system can be upgraded intelligently without installing any hardware, so as to realize the active recognition and early warning of tunnel water accumulation.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
