Detection and Rejection Automation System for tablet medicine packing error

Since medicine is essential for human safety, an improper quantity or broken medication may result in legal action under the Consumer Protection Act. This emphasizes how important accurate packaging is to consumer safety and wellbeing. Therefore, this paper introduces a detection and rejection automation system using computer vision for tablet medicine packing errors. The detection system simultaneously detects the RGB images captured from machine RGB cameras by implementing the traditional computer vision method and a state-of-the-art real-time object detection model, YOLOv8. Pnuemetic with a push arm is used to immediately reject the broken or improper-quality blister pack. A Programmable Logic Controller (PLC) is used alongside the machine hardware to control this automation system. The PLC obtains detection data from computers and controls the rejection and other operations of the machines. PLC and Human Machine Interface (HMI) are integrated to make it simple for individuals to view and control essential information. Our implemented YOLOv8 model achieved an F1-score of 92.3% and an average interface time on a 720p image of 98 ms (using only CPU)