Pavement failure classification (Help!)

Community
1

Hi everyone. My name is Matías, from Uruguay.

I am working to help improve procedures for a government office dedicated to road maintenance. This office has a truck equipped with a camera that can take pictures of the road and georeference each one. However, people are not using this tool because they don’t have any means to analyze the material.

I am trying to find a trained algorithm to detect and classify different types of pavement failures, but I haven’t found anything. I need an algorithm that can classify 20 different types of failures and measure them (area and length).

Do you know anything similar to what I am searching for? I was thinking of an algorithm like YOLO, first training it to detect all types of failures, and then training it again to classify the different types (for the 20 classifications).

Any help would be useful!

Thanks a lot!

2

You might try image embedding. You classify “bad” road, and also “good” road. Your labels are on each of the embedding vectors.

When a new image comes in, you embed it, and check out the top 5 or so images it correlates with. Then you can do a weighted average on the labels, weighted by correlation, to find the most likely label, and get the label standard deviation.

So you will get something like “bad” with low standard deviation, or with high standard deviation. You use both to determine if you can trust the label or not.

This does require extensive labeling. Like 10k-20k images at least. You can knock this out in a week or two though if you work hard enough.