Aleaf: Android-Based Phytotherapy Leaf Recognition Using Custom Vision Machine Learning
Published Jun 13 2022 03:13 AM 2,715 Views
Regular Visitor

Kamusta! I am Althani Miguel G. Leonida, an advocate of tech education, with a passion for Machine Learning and Artificial Intelligence. I am a Gold Microsoft Learn Student Ambassador from the Philippines and currently taking a bachelor's degree in Computer Science specializing in Software Engineering in the Lyceum of the Philippines University-Manila. My passion is to inspire, empower, and help the community through tech and non-tech projects, and so I chose Education and ML + AI as my Leagues in the MLSA program.

 

 

imageArticle.png

 

In the case where I build projects, I create them to provide solutions to communal problems, as I believe this is what research and technology are for. My undergraduate thesis is no different, as the root problem was caused by the COVID-19 pandemic. One of my cousins got into a motorcycle incident which left him with large wounds on his shoulders and knees. Unfortunately, we cannot give him immediate essential medicines as our area is far from the city, and community quarantines were heavily imposed during that time. There were no pharmacies around. What we have are shrubs and trees. Plants that may have medicinal properties. This is one of the factors that helped me push through the idea that our local community now should be able to have the proper knowledge of exercising this kind of healthcare practice - phytotherapy.

 

Application Demo Video: https://youtu.be/8Gb3o0AtbFA

MS Forms Research Instrument (Survey): https://forms.office.com/r/0pf71cGeV2

 

Introduction

Phytotherapy is the practice of using plant-based therapies to treat wounds and illnesses. It comes from the Greek words phyton meaning plants and therapeia denoting treatment thus, “treatment by plants”. Other associated terms with it, in terms of my study,  are herbalism, traditional medicine, indigenous medicine, and ethnomedicine. Even the World Health Organization recognizes its potential, considering phytotherapy as an applicable and accepted treatment method around the globe. But, even with its popularity, correct information about herbal medicine is very nil or poor in our local communities. 

 

Problem

The deficiency of knowledge with regard to the proper administration of herbal medicines exists. This may be due to the fact the practice of phytotherapy is usually based on experience and information is passed down through verbal transmission and there is really no abundance of written documents about it. The proposed leaf recognition Android system will detect whether a plant is medicinal or not through the captured leaf image and patterns. If it detects a medicinal plant, it will display what plant it is, its therapeutic benefits to the human body, the preparation, administration, dosage, and frequency and duration of usage. 

 

function.png

 

Purpose

The community is the first reason why I had taken this study. Because of the limited mobility the people experienced due to lockdowns and community quarantines, the acquisition of essential medicines proved to be really challenging. This proposed system may fill the medical needs of the people as this will help them be informed that a plant just behind their houses may be used as a complementary and alternative medicine to tend their wounds and some of their illnesses, if and only if hospitals and essential medicines are not accessible, which is usually experienced in rural and far-flung areas where even their barangay clinics have a lack of resources.

 

The second entity that may benefit is the clinical pharmacies. Plants offer the possibility of an alternative technique in the discovery of new drugs. This study may also act as a complementary project to the goals of the WHO Traditional Medicine Strategy (2014-2023) and the United Nation’s drug development program.

 

Scope

1. Opted to build the leaf recognition system as an Android application.

2. Focused on phytotherapy for skin diseases.

 

Result 

The first activity is the launching of the application by the user. After, the proposed application will request a leaf image through the camera module and the user will capture an image as a response to this request. If the user denies the image, then the system will ask for a request for the second time around, launching the camera module again. If the user accepts the image, it will be uploaded to the cloud, the Azure Custom Vision service, where it will be analyzed and classified. After a successful analysis, it will return to the Aleaf system and display the necessary information based on the captured image.

 

storyboard.png

From what the Azure Custom Vision metric says, the performance of our image classification model fed with more than 4,000 leaf images, the precision rate is very high. This means that the output to be given by Aleaf is bound to be accurate.

 

 

precision.jpg

 

Before ending this article, I humbly ask our readers to participate in my senior research study and be my respondents. If you are working or studying in fields related to Information Technology, please watch the video demonstration here: https://youtu.be/8Gb3o0AtbFA, and please proceed in answering the 21-item research instrument here: https://forms.office.com/r/0pf71cGeV2. Your response is very much appreciated. Thank you so much!

 

 

Co-Authors
Version history
Last update:
‎Jun 12 2022 09:15 AM
Updated by: