Page 56 - 2025S

UEC Int’l Mini-Conference No.54 49

MPIcam: A Deep Learning Based Real-Time Medicinal Plant

Identification System for Rural Healthcare in Bangladesh

2
1
Shabnur Anonna AKHY ∗1 , Tahir HUSSAIN , Md Bilayet HOSSAIN , and Hayaru
SHOUNO 2
1 UEC Exchange Study Program (JUSST Program)
2 Department of Informatics
The University of Electro-Communications, Tokyo, Japan

Abstract

Plant-based herbal medicine has played a vital role in traditional healthcare in Bangladesh, es-
pecially in rural areas lacking access to modern medical care. However, manual identification of
medicinal plants presents considerable difficulties, as it is time-consuming and usually requires expert
knowledge. To address this issue, we developed the Medical Plant Identification Camera (MPIcam),
a web-based application that enables real-time plant identification through leaf image capture. De-
signed with a simple and user-friendly interface, MPIcam is accessible even to users with minimal
technical experience. The system employs deep learning (DL) techniques based on Convolutional
Neural Networks (CNNs). The dataset, which consisted of 9,150 images from six different categories
of medicinal plant species, was split 80% for training and the remaining 20% equally between testing
and validation. We applied modifications to the suggested methods: VGG16, DenseNet201, Mo-
bileNetV2, Xception, and ResNet50. Based on the evaluation of the models, DenseNet201 achieved
the highest test accuracy of 99.00%, followed by ResNet50 (98.59%), MobileNetV2 (97.99%), Xception
(95.58%), and VGG16 (93.17%). We used the best-performing model in MPIcam with Streamlit to
build a simple and easy-to-use web interface for identifying plants automatically. The proposed sys-
tem offers a practical solution for researchers, herbalists, and the general public to accurately identify
medicinal plants and understand their associated health benefits.

Keywords: Medicinal-Plant Identification, Automated System, Deep Learning
1 Introduction fecting their ability to retain traditional knowl-
edge. This growing dependence on technology
Since ancient times, medicinal plants have been is associated with a decline in human knowl-
a major source of substances used to treat ill- edge about medicinal plants. Thus, there is a
nesses, and it is critical to recognize and classify significant decline in public awareness of tradi-
medicinal herbs for better healthcare [23]. Re- tional herbal remedies, and these practices are
searchers claimed that almost 80% of people are gradually disappearing. However, situations oc-
dependent on herbal plants for their health and casionally arise in which people are temporarily
wellness [23]. In ancient times, people were not disconnected from technology and require medi-
dependent on technology, and advanced medi- cal treatment. Unfortunately, in such cases, ac-
cal systems were not prevalent. Consequently, cess to hospitals, pharmacies, and doctors may
to meet their medical needs, people prepared be unavailable. Basic treatment can be provided
home remedies using various medicinal plants. if knowledge of home remedies using medici-
However, with changing times, people are in- nal plants is available. Besides, many people
creasingly relying on technology, which is af- in modern society do not recognize the correct
medicinal plants nor understand which plants
∗ The author is supported by JASSO Scholarship.