Detection and Identification of Pills using Machine Learning Models

Detection and Identification of Pills using Machine Learning Models

P Siri, Dr. Rishi Sayal

PG Scholar, Department of CSE, Guru Nanak Institutions Technical Campus, Hyderabad.

Professor, Department of CSE, Guru Nanak Institutions Technical Campus, Hyderabad

Abstract: Accurate identification of pharmaceuticals is vital in order to minimize medical errors, as it directly effects the well being of patients and prevents major consequences.Drug abuse puts patients at significant risk for problems and possible injury. Healthcare providers are burdened by this problem since they have to manually search pill databases to find prescriptions when patients are unable to supply their prescription details. Because patients typically throw away the containers that hold their medication along with the prescription, this situation occurs regularly. The development of computerized medication systems that use information technology to precisely identify medications and identify possible interactions between them is imperative in order to address these issues. An inventive deep learning-based pill detection system with sophisticated medicinal substance recognition capabilities is shown in this senior project. The MobileNet architecture serves as the underlying model for the system, which was created with the Python programming language. The primary goal of this research is to deliver intelligent drug identification and accurate pill detection from photos. The system is trained using a dataset of 1,268 samples for both training and testing in order to achieve this. The MobileNet architecture is used in the system's training process, which produces remarkable performance metrics. Both training and validation accuracy are said to have been attained at 98.00%. The system's capacity to precisely identify medications and detect pills is validated by its excellent accuracy rates. This deep learning-based pill identification system has a lot to offer the healthcare industry in real-world applications. It reduces human mistake and saves healthcare personnel important time by automating the pill identification procedure. The system also helps patients by allowing them to get detailed information about their prescriptions and confirm that they are what they were prescribed. Thorough testing on a variety of pill images is part of the system's evaluation process to guarantee its robustness, accuracy, and dependability. The study shows how well the proposed system works to achieve precise pill detection and intelligent medicinal medication identification through a great deal of testing and validation.

Keywords: Pill detection, Image classification, Deep learning, Machine learning, Medication identification, Healthcare automation.