This project presents the development of YoloDefectNET, a deep learning-based model for the detection of defects in printed circuit boards (PCBs). With the rapid advancement of industrial automation, PCB defect detection has become critical to ensuring high product quality. The YoloDefectNET model is designed to enhance the traditional YOLO architecture by integrating multi-scale merging and attention mechanisms. These enhancements allow for more accurate and efficient detection of tiny and complex defects commonly found in industrial manufacturing environments. The model is trained using a dataset of PCB images containing various types of defects, including short, spur, missing holes, mouse bites, open circuits, and spurious copper. The system is designed to operate in real-time with high accuracy, contributing significantly to quality control in PCB manufacturing. The frontend is developed using HTML, CSS, JS for an interactive user interface, while the backend is powered by Python to handle the deep learning models. The integration of attention mechanisms ensures better feature representation, enabling the model to focus on crucial parts of the PCB image. This project aims to offer a reliable solution for defect detection in a production setting, enhancing the speed and accuracy of defect identification, thus reducing human error and improving production quality.

Key Words:

PCB Defect Detection , YoloDefectNET , Deep Learning , YOLO Architecture , Multi-Scale Merging , Attention Mechanism , Industrial Manufacturing , Short Circuit , Spurious Copper