Machine Learning-Based on Abnormality Electronic Circuit Boards Detection System

Electronic circuit boards typically have resistors, capacitors, diodes, and transistors. In the production of electronic circuit boards, there are often problems with parts damaged or lost during production. Due to the failure of the production process, the production standards are not targeted. Most of the electronic circuit boards with defects will be used to inspect, causing fatigue and error, and the failure of these problems can bring computer vision technology to solve the problem. This research focuses on inspecting defective electronic circuit boards with machine learning techniques-the precision and accuracy of detecting the problematic circuit board. By using the CiRA CORE program, which can generate image recognition and in-depth learning algorithms for electronic circuit board inspection, a model with a defect is trained to compare the model using the 2 Models Convolutional Neural Network (CNN) principles, including DarkNet-19 and V4-tiny. Both models' current AVG loss values are 0.0753 and 0.0755, respectively. The accuracy value of the V4-tiny model was 99.75^% more than that of 99.56% of the DarkNet-19 model. Machine learning image classification is performed on the CiRA CORE platform to use the electronic circuit board abnormality detection system, which enables efficient detection of abnormal circuits and reduces fatigue inspection by inspector.