A damage detection system using AI is disclosed. The damage detection system using the AI includes: a field shot image storage unit for storing the divided shot images taken by zooming in a predetermined area over the entire target surface through a camera at a facility inspection site; Damage image storage unit for learning where images of damaged parts of the structure surface are collected and stored in advance, damage confusion image storage unit in which images of construction joints on the surface of structures that can be confused with damage are collected and stored in advance, structures without damage A deep learning module including an intact image storage unit in which surface images are collected and stored in advance, and an artificial neural network for learning based on the images stored in each storage unit so that damage detection performance is established; An image merging module for merging the split shot images stored in the field shot image storage unit to realize a large image in which the entire target surface is displayed; And a control unit for executing the deep learning module and the image merging module, and updating the split shot image captured at the facility inspection site to the field shot image storage unit; Substituting the split shot images updated in the field shot image storage unit into the deep learning module, and executing the deep learning module to detect a damaged part in each split shot image; Executing the image merging module to perform the step of merging each segmented image after detection of the damaged part is completed to implement a large image in which the entire target surface is displayed, and displaying the detected damaged part on the large image Characterized in that.
展开▼
