A Machine Learning-Based Approach for the Detection of Drugs in Drug Self

Abstract

Accurate detection of specific named medicines on drugstore shelvesis critical for pharmaceutical inventory management. An automatic medicine name and box detection system can save time and hassle for the pharmacy owners and the customers. Existing research did not properly use machine learning technology to accurately detect medicine from drug samples. This paper aims to create a machine learning-based medicine detection system capable of automatically recognizing and localizing medicine boxes on shelves. The system uses Faster region-based convolutional neural network (R-CNN) and YOLOv5 to recognize and extract medicine boxes from images. Text recognition techniques, such as Tesseract OCR, are used to extract medicine names from boxes. This work includes collecting and annotating a dataset, training and evaluating models, and implementing text recognition algorithms. The simulation results show that the proposed faster R-CNN and Tesseractbased system is more accurate at detecting medicine boxes and extracting text than the existing Yolov5 and Tesseract-based systems.

Received: 23 February 2024 | Revised: 2 August 2024| Accepted: 31 October 2024

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

The GitHub data that support the findings of this study are openly available in https://github.com/argman/EAST. The GitHub data that support the findings of this study are openly available in https://github.com/clovaai/CRAFTpytorch. The GitHub data that support the findings of this study are openly available in https://github.com/bgshih/crnn. The GitHub data that support the findings of this study are openly available in https://github.com/MhLiao/TextBoxes. The GitHub data that support the findings of this study are openly available in https://tesseractocr.github.io/tessdoc/. The GitHub data that support the findings of this study are openly available in https://github.com/tesseract-ocr/tesseract.

Author Contribution Statement

Md. Nazmul Sakib: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Resources, Data curation, Visualization. Mahfuzulhoq Chowdhury: Conceptualization, Validation, Formal analysis, Investigation, Writing – original draft, Writing – review & editing, Supervision, Project administration.