The invention discloses an image-based power equipment identification method. The method comprises the steps of receiving an electrical equipment image training set, performing preprocessing, establishing image feature point description information, association point information and indexes of the image feature point description information, the association point information and electrical equipment categories for preprocessed images by using an image recognition algorithm, storing the image feature point description information, the association point information and the indexes in a feature library, and training the preprocessed electrical equipment image training set by using machine learning; Receiving a to-be-identified power equipment image, performing preprocessing, extracting feature points, and performing image search based on a feature point detection SIFT algorithm; Establishing a feature set index for the feature points extracted from the to-be-identified power equipment image, matching the feature set index with the index of the feature library, and identifying the type of the power equipment; And carrying out difference and cumulative image processing on the area wherethe power equipment is located in the power equipment image, comparing with the normal image of the power equipment of the corresponding category, judging the operation condition of the power equipment, and obtaining the fault or the fault hidden danger of the power equipment.