Method and equipment for detecting pitch of adjacent racks of rack rail railway

The invention relates to the field of mountain rail traffic rack rails, in particular to a rack rail adjacent rack pitch detection method and equipment. According to the method, the rack rail joint is mechanically learned by using the neural network model and the pre-collected rack rail image data, so that the convolutional defect detection network model capable of quickly judging whether the rack rail joint is qualified or not is constructed, and the detection accuracy and reliability of the rack rail joint are greatly improved; meanwhile, due to the full-intelligent detection process, the detection efficiency is also greatly improved, and the labor cost of detection is also reduced; and meanwhile, by numbering the to-be-detected toothed rail railway, the position of the unqualified toothed rail joint and the corresponding line information can be output at the first time when the toothed rail joint is found to be unqualified, so that a worker can find and process the unqualified toothed rail joint at the fir