Manhole Cover Detection Using Depth Information

Abstract

Nowadays, the security of manhole cover has become a growing concern. In the past, manual inspection and Internet of Things were often used to ensure the safety of manhole cover. With the advent of smart cities, people look forward to using cameras as a more convenient method to detect whether manhole covers are damaged. Based on the structure of object detection network, we integrate the depth information in the picture and create a new model which improves the accuracy of detecting damaged manhole covers.