The invention relates to an artificial intelligence detection and diagnosis method for visible light imaging big data of power transmission equipment, and the method employs a five-layer deep convolution neural network to detect the background, tower and insulator regions in a visible light image of a power transmission line. A power transmission line data set containing multiple types of targetsis constructed, wherein the data set comprises a background, a tower, a glass insulator and a composite insulator. The network is pre-trained by using the Cifra-100 data set, and then network optimization is performed by using the power transmission line data set. During detection, an image is divided into image blocks, the category of each image block is identified by using a neural network, andthe areas where the image blocks of the same category are located are merged to obtain the area of the category of target in the image. It is verified that the method is obviously superior to a traditional method, the method can be used for typical target detection in a visible light image of a power transmission line, and a detection result can be used for diagnostic reference or further targetstate analysis.