Research on identification and location of blocked ore at ore bin inlet based on Mask RCNN

When the inlet of the ore bin is blocked, there are complex site conditions and difficulty in ore identification and detection in the process of ore identification. To solve these problems, the research on intelligent identification and detection of ore was carried out by using deep learning and image processing techniques. The method of ore identification and detection based on Mask RCNN was proposed, which could realize instance segmentation while identifying ore. It was proposed to replace the circumscribed rectangular frame positioning method in Mask RCNN with the centroid coordinates of the ore contour, effectively solving the problem of inaccurate ore positioning. The experimental results showed that the ore recognition model based on the Mask RCNN network could accurately achieve ores of multiple numbers, different positions, and stacked, with a comprehensive accuracy rate of 97.6%. The positioning method using the coordinates of the ore contour and centroid could effectively avoid positioning errors caused by shapes and positions of ores, providing precise visual guidance for the intelligent blocking manipulator.