Multi-Class Fruit Leaf Classification Dataset (10 Classes)
Description
This dataset comprises 3,173 high-quality images of healthy fruit leaves from 10 different classes, specifically curated for research in plant classification, species identification, and agricultural analysis using deep learning and computer vision techniques. The dataset includes images of Aegle marmelos (336), Black plum (304), Custard Apple (304), Guava (325), Jackfruit (311), Lotkon (306), Lychee (312), Mango (330), Plum (302), and Star Fruit (343). Each image was captured under diverse environmental conditions to ensure a robust dataset for training and evaluating machine learning models. Since all images represent healthy leaves, this dataset can serve as a baseline for plant disease detection, enabling researchers to compare healthy and diseased samples effectively. It is well-suited for image classification, feature extraction, transfer learning, and species recognition in the fields of agriculture and botany. Potential applications include training convolutional neural networks (CNNs) and transformer-based models for fruit leaf classification, fine-tuning pre-trained models, and developing AI-driven plant monitoring and smart agriculture solutions. The dataset also serves as a valuable resource for augmenting existing datasets to improve model generalization. Researchers and AI practitioners can leverage this dataset to advance precision agriculture and plant health monitoring. For any inquiries or collaboration, please contact the authors.
Files
Steps to reproduce
The Multi-Class Fruit Leaf Classification Dataset (10 Classes) was collected from nursery gardens, where only healthy fruit leaves were photographed under good environmental conditions. Images were captured using Realme 7 Pro and Realme 8 Pro mobile phones, ensuring high resolution suitable for machine learning tasks. The collection process involved selecting fresh and undamaged leaves from trees, with species identification verified through expert consultation and botanical references. Photos were taken under natural daylight. Leaves were primarily placed on a plain white background to enhance clarity and reduce noise. After capturing the images, they were manually reviewed, categorized, and organized into labeled folders for each fruit species. Quality control steps included removing blurry or misclassified images and standardizing resolution while maintaining high quality. This dataset provides a high-quality and well-structured collection of healthy fruit leaves, making it suitable for species classification, feature extraction, transfer learning, and AI-driven agricultural research. The outlined methodology ensures reproducibility, allowing researchers to expand the dataset by following similar data collection and processing techniques.