|
Title:
|
HMEDASYS: HOME MEDICATION INTAKE ALERT SYSTEM |
|
Author(s):
|
Jardson Rodrigues, Christophe Soares, Pedro Sobral, Rui S. Moreira and José M. Torres |
|
ISBN:
|
978-989-8704-44-3 |
|
Editors:
|
Hans Weghorn and Pedro Isaias |
|
Year:
|
2022 |
|
Edition:
|
Single |
|
Keywords:
|
Drug Recognition, Medicine Recognition, Deep-Learning, Computer Vision, YOLO |
|
Type:
|
Full Paper |
|
First Page:
|
148 |
|
Last Page:
|
155 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
The use of medication as treatment is often crucial to improve a patient's health. This process involves several factors, and often patients forgot to take their medicine. Thus, mechanisms that can contribute to therapy adherence are fundamental. The goal of this work has been the development of a real-time and automatic video object recognition system (HMEDASYS) to be used in the context of adherence and medication taking in-home care. The methodology followed has been to train a classifier model for medication pack recognition using Deep Learning. Combining and organizing data from previously annotated images, we retrained a neural network using the YoloV4-tiny Framework. In the experiments developed, two different classes of medicines were used for training and the system achieved a mAP of 98.33% on the test set. During the operation of HMEDASYS, the medication packs are real-time monitored using a video camera coupled with the developed system. In further 8 evaluation scenarios of the operation, only two errors were verified in a total of 16 medication packs to be identified. |
|
|
|
|
|
|
