Deep Learning for Classification and Efficacy of Medicinal Plants for Managing Respiratory Disorders
Year: 2025
Type: conference-paper
Abstract: Respiratory diseases present significant health challenges globally, prompting research into non-pharmaceutical alternative treatments. Medicinal plants have been extensively utilized in traditional medicine due to their potential therapeutic benefits. These plants contain bioactive compounds with expectorant, antimicrobial, and anti-inflammatory properties that can help alleviate throat irritation, reduce congestion, and relieve coughing. Phytopharmaceuticals, generally low in toxicity and cost-effective, are often derived from medicinal plants undergoing scientific research and standardization. While numerous studies demonstrate the effectiveness of medicinal plants in treating respiratory conditions, factors like small sample sizes, variations in herbal preparations, and a lack of standardization hinder the generalizability of these findings. As interest in naturopathy grows as an alternative to allopathic medicine, many individuals turn to herbal remedies. However, caution is warranted, as herbal remedies may pose potential side effects, safety concerns, and interactions with prescription medications. This research aims to investigate the efficacy of medicinal plants in treating respiratory ailments based on scientific literature. Additionally, the study explores the classification of medicinal plants using deep learning models such as DenseNet 121, Inception V3, and EfficientNet B4 applied to a dataset fetched from Mendeley, which includes images of 80 different leaf classes of medicinal plants. The proposed enhanced EfficientNet B4 model outperformed the other models, achieving a testing accuracy of 91.37%, precision of 91.97%, recall of 91.37%, and an F1 score of 91.40%. The results suggest that our given model performs effectively on the provided dataset.
Source: Procedia Computer Science
Authors: Samia Khan, Farheen Siddiqui, Mohd. Abdul Ahad
Institution: Jamia Hamdard
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4410253003