Medication Recording System Using Oral Image Acquisition and Image Recognition

Abstract

In the aging society, especially those with eye diseases, often forget to take medicine or take wrong medicine. In the new drug approval process, objective and accurate medication records have become very important. To improve the reliability of medication recording, we have been developing a cup-shaped medication recording device by taking intraoral pictures. This device has a camera with a wireless transmitter and a rechargeable battery in the cup. To recognize oral medicine accurately, we used AI image recognition technology. We selected four representative drugs (500 images for each) for testing, and the recognition accuracy can reach more than 90%. While ensuring the recognition accuracy, we achieved real-time detection, which can reach the detection speed of 16 frames per second on smartphones. In addition, an Android application based on AI image recognition has been developed that is easier to use. The medication management application was equipped with image acquisition, image recognition, medication recording, forgetting of medication, and caution for wrong medication.