CIM manhole cover state visual inspection system based on deep learning and multi-view

Abstract

The invention discloses a CIM manhole cover state visual detection system based on deep learning and multiple viewing angles. The system realizes the manhole cover state visual detection based on the manhole cover city information model MCCIM. Attribute information and geographic location information of nearby traffic light signals, the system includes a perception unit, a perspective transformation unit, a manhole cover positioning unit, a manhole cover state judgment unit, a traffic light control unit, and a visual management pipe unit. The invention reduces the cost required by the system in the state detection of the manhole cover, improves the accuracy of judging the state of the manhole cover, and improves the safety of urban traffic by controlling the traffic lights near the road where the manhole cover is missing or displaced.

Description

technical field [0001] The invention relates to the technical field of smart cities, in particular to a CIM manhole cover state visual detection system based on deep learning and multiple viewing angles. Background technique [0002] In recent years, due to rampant lawbreakers, manhole covers on urban roads have been lost frequently. This has serious implications for road safety. Especially at night, when the ambient light is dim, vehicles and pedestrians passing normally on the road are prone to safety accidents because the lack of manhole covers cannot be observed. Due to the non-standard construction management of some units, the manhole cover may also shift, forming a hidden danger that is not easy to observe. Therefore, the missing and damaged manhole cover poses a huge safety hazard to normal driving. [0003] Some solutions use multiple sensors installed at the bottom of the manhole cover to monitor the state of the manhole cover. On the one hand, this method requ...

AI Technical Summary

Problems solved by technology

On the one hand, this method requires additional power supply for the sensor. On the other hand, the sensor is easily affected by the environment and the cost is high, which is not suitable for large-scale application in cities.

[0004] Some solutions use a single camera to detect or locate the manhole cover. On the one hand, a single camera is easily blocked by trees, birds, etc., resulting in detection failure.

On the other hand, the data of a single camera is too single, and it is easy to cause misjudgment of the manhole cover

[0005] Therefore, the existing manhole cover state detection technology has the problems of high cost and easy misjudgment

Images