Project description:

    At present, a single device can produce 200K to 300K LED lamp beads per hour, and the core product is 1 to 3MM in size, so it's impossible to conduct 100% appearance detection, and only a small number of samples can be selected for manual visual detection according to the production batches. The detection results can barely be used for reference, which can not meet downstream customers' strict requirements for product quality. Based on Handdle AI algorithm platform, DSTEK can accurately identify and locate the types and locations of product defects through machine vision and AI algorithms.

 Defect types:

More glue on the surface, less glue on the surface, bubbles on the glue surface, abnormal glue surface, cracks on the glue surface, scratches on the glue surface, etc. There are more than 15 conventional detection projects in total; Reverse side: pin glue, pin deformation, pin absence, etc.

                                                                                                                          OK  sample                                                                                                      Detection  picture

          AI algorithms are 50 to 80 times faster than the traditional algorithms          Overall intelligent detection and a 60% cost reduction