The project aims to reduce capital and time losses incurred due to faulty or unfinished PCBs by providing an accuracy of 75.48% in fault detection. The system utilizes various techniques, including image processing, machine learning, and in-circuit testing, to detect and address issues such as track errors, missing components, component orientation, and other manufacturing or assembly-related problems.
The use of CNN in this project enables automated classification and fault detection in PCBs, allowing for the identification of various types of faults, such as missing components and misorientation of components. The system's future scope includes optimizing the algorithm and extending its capabilities to inspect single-layer PCB faults and other advanced error detection techniques.
Runtime accuracy of the model
Training and validation graph
Picture accuracy
This project is used by the following companies: