MEP: A Comprehensive Medicines Extraction System on Prescriptions

Accurately identifying drug names from prescription images is essential for effectively processing and managing patient medical information. In order to enhance our previous approach, we have introduced a new version named Medicines Extraction on Prescription (MEP). This innovative approach employs heuristic rules and a Temporal Convolutional Network model to extract and classify proprietary medicines from prescription images captured by smartphones. Our model not only achieves a precision score of up to 0.94 on experimental datasets, but it also boasts impressive processing speed. The total drug name recognition and average extraction time is only 6.67 s per prescription, significantly faster than the 17.81 s average processing time of our previous model.