Cap-Eye-citor: A Machine Vision Inference Approach of Capacitor Detection for PCB Automatic Optical Inspection

• Published in: 2020 IEEE 7th International Conference on Engineering Technologies and Applied Sciences (ICETAS) pp. 1 - 5
• Main Authors: Susa, Julie Ann B., Mariquina, Erwin, Tria, Meriam L., Adolfo, Cid Mathew, Castro, Julius C.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 18.12.2020
• Subjects: Automatic Optical Inspection; Capacitor; Capacitors; Circuit Boards; Object detection; Optical detectors; Optical imaging; Printed circuits; Production; Software algorithms; YOLO V3 Algorithm
• DOI: 10.1109/ICETAS51660.2020.9484182

Circuit boards are one of the key elements of the electronics industry. Because of its demand for portable electronic goods, the production of circuit boards has become more significant. The most significant aspect of printed circuit board production is AOI or Automatic Optical Inspection. When generated in a single quantity, the PCB requires a small way to teach and change the AOI method for process validation. This paper proposes a mechanism of detection of capacitors trained on circuit boards using the YOLO V3 algorithm. YOLO is a form of rapid object detection based on the convolutional neural network or CNN. CNN's deep network can distinguish specific characteristics from all the image features. The study developed an AI with the same feature that the manufacturing industry uses to assist students with an effective evaluation of their circuit boards. It focuses on the capacitor in a circuit board. It will not, however, be capable of distinguishing other forms of electronic components. Nearly 60 percent of the entire test has a performance of 100 percent detection, according to the test results, while 33.33 percent is over 70 percent detection and 6.66 percent has a 0 percent detection output. Capacitor recognition has been done with an overall testing accuracy of 93.33%.