Medicine Classification Using YOLOv4 and Tesseract OCR
Charmaine C. Paglinawan, Marielle Hannah Caliolio, Joshua B. Frias
Abstract
For seniors, consuming the incorrect pharmaceutical drug is a common but serious medical issue. Poor vision, difficulty reading, impaired memory, and problems with fine-motor skills are all factors that might lead to using the incorrect pharmaceutical drug. Previous studies have created systems that can automatically identify medications to address this issue. Although, this research lacked the integration of the latest official version of YOLO and the Tesseract OCR. As such, the researchers developed a system that can classify medicines using YOLOv4 and Tesseract OCR. Through tests, YOLOv4 was established to be the best in accuracy (~95% accuracy), followed by YOLOv4 and Tesseract OCR together (~19% accuracy). The least accurate one was using the Tesseract OCR only (~14% accuracy).