Two-stage workpiece recognition algorithm based on Mask-RCNN and template matching

To achieve automated sorting and transportation of workpieces in the factory, AMR is used to assist in workpiece handling. This paper proposes a two-stage Workpiece Recognition Algorithm based on Mask-RCNN and template matching for determining the processing stage of workpieces. First, Mask R-CNN is used for instance segmentation and reverse masking to obtain an image of the target workpiece, which serves as the image to be matched. Relying on the three-view angle of the workpiece's external shape during the processing stage, a template image is created, and SIFT-FLANN is used for image matching to identify the processing stage. Finally, experiments show that the method achieves an accuracy rate of 81.27% for recognizing categories of workpieces, with an average matching time is close to 20ms for identifying process stages of disk and sleeve-type workpieces, and a matching accuracy of 98.75%, allows for accurate determination of workpiece type and processing stage. This provides a basis for AMR to transport workpieces and contributes to improving factory production efficiency and reducing production costs.