| Summary: | The detection of defects in a product is one of required production process for quality control. Currently, the quality control process of metal screws uses many manpower for manual inspection. Hence, this study about to implement faster region-based convolutional neural network (faster R-CNN) to detect the micro-defects on metal screw surfaces. The defects of surface damage, stripped screw, and dirty surface screw considered in this research. Raspberry Pi 3 with a camera module is used for image acquisition of the metal screws in determining various kinds of defects. The image is also acquired to be used for the training of the faster R-CNN. A testing is carried out to test the performance of the model. The experiment outcome shows that the detection accuracy of the model is 98.8%. The model also shows superiority in this project detection method compared with the traditional template-matching method and single-shot detector (SSD) model. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.
|
|---|