Efficient Detection of Missing and Misplaced Components in Electronic Boards through Edge Impulse’s Automated Inspection

The detection of missing and misplaced components on electronic boards is a critical and important task in the quality assurance process of electronic manufacturing. Traditional manual inspection methods are prone to human errors and are time consuming. In this work, we proposed an automated approach using Edge Impulse’s transfer learning technique. We trained a neural network model MobileNetV2 using transfer learning and evaluated the model’s performance on a test set. The results show that our trained model have achieved an accuracy of 94.2% validation set of trained data set and 96.55% misplaced components. We trained our model with twelve possible cases of errors and demonstrate the practical applicability of our approach by deploying the model on a Raspberry Pi 4 using Edge Impulse’s deployment tools. Furthermore, the use of Edge Impulse’s transfer learning and deployment tools makes our approach cost-effective, scalable, and easy to implement in real-world manufacturing environments.