Inspection apparatus and method based on deep learning

摘要

A facility defect detection apparatus and method using deep learning are disclosed. A facility defect detection apparatus using deep learning includes: a line scan camera for photographing a facility image; An image reconstruction unit for matching the facility image and the reference image using a phase correlation result between the facility image and the reference image; A facility detector configured to detect a facility by applying the matched facility image and the reference image to a deep learning-based facility detection model; And a transformation inspection unit that performs image conversion through homography and light flow estimation between the two facilities detected in the facility image and the reference image, and detects a defect area by removing noise through a morphology operation on the two facilities converted into the image. Includes.