Graphical user interface (GUI) refers to the interface that is convenient for users to operate the computer through graphics to realize the communication between users and computers. The object detection and image segmentation methods based on deep learning used in this paper rely on the python interface provided by Caffe. The contour feature extraction and wear degree judgment rely on the rich library functions provided by python, and the QT powerful GUI programming library has been integrated on python. Therefore, this paper uses Python 2 The visual interface of bucket tooth fault detection program is designed by PyQt4 in 7 environment.
According to the functional requirements of the bucket tooth system, the user interface shown in Figure 1 is designed. The input image area in the main interface is used to display the detected image; The target detection area is used to display the result of target detection, and directly mark the detected bucket teeth on the image through the frame; The image segmentation area is used to display the image segmentation results of the detected bucket teeth and the faulty bucket teeth; The detection information is to visually display the specific fault analysis through numbers, including the number of bucket teeth in the template image, the number of bucket teeth in the detected image, the number of fault bucket teeth, the position of fault bucket teeth and the wear degree of fault bucket teeth.
The initialization key is to assist the user to complete the initialization setting, as shown in Figure 2. When the initialization key is pressed, the user specifies the number of template bucket teeth, the saving path of the detected image and the storage path of the detection results. The specified content can be displayed and returned to the graphical interface program as a parameter. After the user completes initialization, click the detection key to start detecting the image file under the corresponding path. As shown in Figure 3, the detection results are displayed in the corresponding area through the interface.
During the operation of the program, if the number of target detection matches, the program will automatically detect the next image. When the target detection program detects that the number of bucket teeth does not match, a warning message will pop up, and the subsequent analysis program will analyze the wear degree of bucket teeth, display the results on the interface, as shown in Figure 4, and prompt the user for subsequent processing.
When the user has processed the fault bucket or ignored the alarm, click the Continue button to let the program detect the next image.