Railway rock fall detection have what good method? Now I can think of is to extract tracks, then using differential extract tracks in the middle of the target goals (if any), and then by examining whether rockfalls, detection method can be use a variety of operator or using the ideas of classification, the second might think of is to realize the detection and classification of direct use of ideas about the whole picture for detection and classification, can be finished with deep learning, but the two have difficulty and limitations, the first kind of track detection itself is very troublesome, haven't seen any good method can extract all kinds of conditions of track, testing may also have the impact because of the shadow, the second is mainly to full operation, or similar to the first combination of steel rail operation objects too much, the probability of false detection should be bigger,

Philosophy can help great god is there any good method or? Be urgent!

CodePudding user response:

Directly to a motion detection, line not line

CodePudding user response:

Collect such as at least 5000 pictures of different railway rockslides and not rock fall, training neural networks,