Intelligent track image analysis system

Abstract

The invention discloses a track image intelligent analysis system which comprises a track image automatic classification module, a track surface damage intelligent identification module and a model iteration updating module. The track image automatic classification module is used for classifying an input image classification model through deep learning into a ballast track, a ballastless track, a turnout area and a pinch plate type fastener track; the track surface damage intelligent identification module is used for selecting an area to be detected after image classification is completed, calling the trained deep learning target detection model of the corresponding track area, identifying the selected area, and extracting and marking a damage interval; and the model iteration updating module is used for performing iteration training on the deep learning target detection model. Collected track image data can be detected through the deep learning model, and the functions of automatic classification of track images, intelligent identification of track surface damage, collection and retraining of damage images and iterative updating of the deep learning model are achieved.

Description

technical field [0001] The invention relates to the technical field of rail safety, in particular to a rail image intelligent analysis system. Background technique [0002] Under the load of temperature, train braking and concrete shrinkage, there will be many damages on the track surface, and it is necessary to arrange remediation in time to avoid potential safety hazards. The existing ones are to take pictures of the tracks by guards and line patrollers, and then manually determine the damage zone. This method is time-consuming, labor-intensive, and highly subjective, which is not conducive to data collation and subsequent analysis of the entire life cycle. Therefore, an orbit image intelligent analysis system is designed. SUMMARY OF THE INVENTION [0003] In view of the above-mentioned technical deficiencies, the purpose of the present invention is to provide an orbit image intelligent analysis system, which detects the collected orbit image data through a deep learni...

AI Technical Summary

Problems solved by technology

This method is time-consuming, labor-intensive and highly subjective, which is not conducive to the whole life cycle data collation and subsequent analysis

Images