Medicinal Leaf Dataset

****Overview of Medicinal Plant Dataset and Its Significance
Mother Earth thrives with an abundance of plant life, many of which play a vital role in health and wellness. These plants serve as essential resources for drug development, herbal product formulation, and treatments for a wide range of ailments. For over 5,000 years, Ayurveda, India’s ancient system of medicine has remained a respected and widely practiced tradition. India, in particular, is home to a rich diversity of medicinal flora.
Various plant parts including leaves, bark, roots, seeds, and fruits are commonly used in the preparation of herbal remedies. These natural medicines are increasingly favoured in both developing and developed nations as alternatives to synthetic drugs, largely due to their minimal side effects.
****Need for Technological Intervention
Identifying medicinal plants by visual inspection alone can be challenging, time-intensive, and prone to error. With many species facing extinction according to IUCN records, leveraging image processing and computer vision technologies becomes essential for accurate identification and conservation. Digitizing medicinal plants is a critical step toward preserving biodiversity.
****Dataset Composition and Collection
To support intelligent recognition systems, a robust dataset of medicinal plant leaves has been curated. This dataset includes 1,010 high-resolution images representing 9 species, with each species contributing between 100 to 120 images. Examples include Piper betle (Betel), Trigonella foenum-graecum (Fenugreek), Bergera koenigii (syn. Murraya koenigii) (Curry Tree), Mentha spp. (e.g., Mentha spicata, Mentha × piperita) (Mint), Senna tora (syn. Cassia tora) (Tora Leaves), Coriandrum sativum (Coriander), Allium cepam (Onion), Raphanus sativus (Radish), Aloe vera (syn. Aloe barbadensis) (Aloe Vera), among others. Each folder is labelled with the plant’s scientific name for clarity.
Leaves were carefully collected from different plants of the same species found in local gardens, ensuring minimal disruption to the environment. Only healthy, mature leaves were selected, and efforts were made to avoid unnecessary waste. Images were captured using a Nothing mobile of 50 MP OIS camera. To enhance model training, leaf images were slightly rotated and tilted to introduce variability.
Impact on AI Research
This medicinal plant leaf dataset is a valuable resource for developing machine learning and deep learning models. It enables researchers and computer scientists to identify plant species, diagnose diseases, and explore the therapeutic properties of herbs. By making this dataset publicly available, we aim to catalyse research in the field of medicinal botany, addressing the current gap in accessible datasets and fostering innovation in plant-based healthcare solutions.
* Ph.D. work of Ms. Amruta Jadhav
* Collaborative work of Government College of Engineering, Aurangabad and JSPM University Pune

## 药用植物数据集概述及其研究意义
地球孕育了种类繁多的植物类群，其中诸多物种在健康养护领域发挥着不可或缺的核心作用。这些植物是药物研发、草本产品配方以及多种疾病治疗的关键资源。印度传统医学体系阿育吠陀（Ayurveda）已有五千余年的发展历史，至今仍是备受尊崇且广泛应用的医疗传统。印度本土拥有极为丰富的药用植物多样性。

通常可用于制备草药方剂的植物部位涵盖叶片、树皮、根部、种子与果实等。这类天然药物因副作用极低，在发展中国家与发达国家均日益成为合成药物的优选替代方案。

## 技术干预的必要性
仅依靠目视鉴别药用植物往往极具挑战性，耗时耗力且易出现差错。据国际自然保护联盟（IUCN）记录，诸多药用植物物种正面临灭绝威胁，因此借助图像处理与计算机视觉技术实现精准识别与物种保护已成为当务之急。将药用植物数字化是保护生物多样性的关键举措。

## 数据集构成与采集流程
为支撑智能识别系统的研发，本团队精心构建了一套高质量药用植物叶片数据集。该数据集包含1010张高分辨率图像，涵盖9个物种，每个物种的图像数量介于100至120张之间。示例物种包括：蒌叶（Piper betle）、胡芦巴（Trigonella foenum-graecum）、咖喱树（Bergera koenigii，异名Murraya koenigii）、薄荷属植物（Mentha spp.，如留兰香Mentha spicata、胡椒薄荷Mentha × piperita）、望江南（Senna tora，异名Cassia tora）、芫荽（Coriandrum sativum）、洋葱（Allium cepa）、萝卜（Raphanus sativus）、芦荟（Aloe vera，异名Aloe barbadensis）等。为便于区分，每个文件夹均以植物的学名作为标签。

叶片样本均采集自当地园区内同一物种的不同植株，尽可能减少对生态环境的干扰。仅选取健康成熟的叶片，同时避免不必要的资源浪费。图像使用搭载5000万像素光学防抖（OIS）摄像头的Nothing手机拍摄完成。为提升模型训练效果，研究人员对叶片图像进行了轻度旋转与倾斜操作，以增加数据多样性。

## 对人工智能研究的价值
本药用植物叶片数据集是开发机器学习与深度学习模型的宝贵资源。它可帮助研究人员与计算机科学家实现植物物种识别、病害诊断以及草本植物治疗特性的探索。通过公开该数据集，我们旨在推动药用植物学领域的研究进展，填补当前可公开访问数据集的空白，并促进植物源性医疗解决方案的创新研发。

* 阿姆鲁塔·贾德哈夫（Amruta Jadhav）女士的博士研究成果
* 奥兰加巴德政府工程学院与浦那JSPM大学联合合作项目