Miss. Kiran Ravindra Manole.

https://doi.org/10.64823/ijter.2606015

This paper presents the design, development, and performance evaluation of a Smart Job Distribution and Quality Verification System built for small-to-medium manufacturing environments. The system merges RFID-based operator authentication, microcontroller-driven conveyor control, real-time ultrasonic detection, USB camera-based image acquisition, OpenCV classical inspection, and YOLOv5 deep-learning defect detection into a single, cohesive platform. A private dataset named Job QC Dataset, comprising 640×480 JPEG images annotated with Label Image and split 70/15/15 for training, validation, and testing, was used to train the YOLO model on Google Colab. Performance metrics including precision, recall, F1 score, and confusion matrix are reported. The system achieved a mean Average Precision (mAP@0.5) of 91.3%, with a precision of 0.934, recall of 0.887, and an F1 score of 0.910 on the test partition. These results confirm the viability of the proposed hybrid inspection framework for industrial deployment.