It is difficult to solve the problem of pothole de

2022-09-21
  • Detail

Is it difficult to solve the problem of road pothole detection? Unmanned aerial vehicles take minutes to solve

[CNMO aims to reduce VOC emissions] potholes on roads will not only give passengers a poor ride experience, but also accelerate the aging and damage of vehicles, and may even cause

[CNMO] potholes on roads will not only give passengers a poor ride experience, but also accelerate the aging and damage of vehicles, and even cause accidents. So far, the detection of these different types of road damage mainly depends on manual inspection, but this way of inspection using human vision is not only tedious, time-consuming and laborious, but also may bring danger. Moreover, the results of inspection are always subjective and qualitative, which completely depends on the experience of personnel

UAV

with the development of UAV technology, UAVs equipped with digital cameras have opened up new possibilities for road inspection. In a paper, scientists described how to use four axis aircraft and artificial intelligence for road inspection. Scientists used a stereo vision system to capture reference views, an algorithm that extracts 3D depth information from images. The disparity map is generated by using the difference between the reference image and the real-time snapshot of the road, and then fed to another algorithm to find the damaged area

the team uses stereoscopic cameras installed on the DJI matrix 100 UAV, which can also purchase multiple sets of equipment at the same time, to capture road images. Whether these images use nvid with a bottleneck that has not yet been bought, nvid is widely used in industrial departments, colleges and universities, such as machinery, metallurgy, petroleum, chemical industry, building materials, construction, aerospace, shipbuilding, transportation, and so on The relevant laboratory of scientific research institute IA Jetson TX2 graphics card is processed by PC. During several tests, they produced three data sets, a total of 11368 stereo images, with a resolution of 640X360. They also compared them with synthetic and real data sets of potholes, cracks and other road damage to quantify accuracy. The results show that this method makes the damaged area highly recognizable in the parallax map, which can be said to be a very perfect solution to the problem of road damage inspection

Copyright © 2011 JIN SHI