Abstract:The states (on/off) of plates and indicator lights are the most direct reflection of the operating status of secondary equipment, and the identification of them is of utmost importance during the intelligent inspection process of secondary equipment. Based on machine vision technology, this paper proposes corresponding image recognition algorithms according to different characteristics of the plates and the indicator lights. A recognition algorithm based on HOG features and SVM classifier is proposed for the plates, the HOG features of the plates are extracted and input into the SVM classifier, and then, the recognition results are obtained by classifying the status of each target plate. A state recognition algorithm for indicator lights based on an improved LeNet-5 network is proposed, which utilizes computer deep learning to achieve the recognition of the states of indicator lights. The experimental results show that the recognition accuracy of the above two algorithms can reach more than 98%.