Artificial intelligence defect recognition system is an image recognition method based on in-depth learning convolutional neural network and wavelet moment, which can quickly identify and analyze the data of machine patrol and human patrol, support the majority of channel hidden dangers in transmission channel and defects , such as pole and tower body parts of metal tools, insulators, shock hammer ,pins,etc.

√ Among a large number of original images of power inspection, it can automatically analyze and distinguish defects , at least in 5 categories and 30 categories.

√ Successfully passed the algorithm verification of State Grid Transportation Inspection Department, algorithm integration of control platform， artificial intelligence identification and verification of State Grid General Aviation helicopter patrol inspection images.

√ The average processing speed is 2 pieces/second, which further reduces the human input in the process of data analysis and improves the intelligence level of data analysis.