Combined Citrus maxima (Pomelo) and Phyllanthus emblica (Indian Gooseberry) Leaf Dataset An Expert-Validated Balanced Resource for Deep Learning-based Disease Detection in Bangladesh

This dataset is a comprehensive collection of Pomelo (Citrus maxima) and Indian Gooseberry (Phyllanthus emblica) leaf images, developed at Daffodil International University, Dhaka, Bangladesh, to support research in machine learning, deep learning, and computer vision–based plant disease detection. A total of 4,502 images (comprising 868 original field-captured samples and 3,634 augmented counterparts) were collected from homestead agricultural gardens in Tangail, Bangladesh, under natural lighting conditions between January 13, 2025, and October 20, 2025. All images were captured using a Realme 7i smartphone (64MP) and standardized to a resolution of 224 × 224 pixels in RGB format. The dataset covers nine distinct classes across two crops: Pomelo (Fungal Curling, Yellow Mosaic, Leaf Hole, Chewing Damage, Healthy, and Early Stage) and Indian Gooseberry (Healthy, Partially Defoliated, and Early Stage). The inclusion of Phyllanthus emblica serves as a morphological negative control for cross-species validation. To address initial class imbalance and improve model generalization, extensive data augmentation techniques—such as rotation, flipping, zooming, shearing, and brightness adjustment—were applied. The dataset's technical utility was validated using state-of-the-art models like Xception, ResNet152V2, and DenseNet121, with DenseNet121 achieving a superior accuracy of 99.00%. This dataset serves as a valuable resource for advancing research in precision agriculture, smart farming, and AI-based crop health monitoring, and is freely available for academic and research purposes with proper acknowledgment to the creators and Daffodil International University.