Agri-Vision4: A Comprehensive Multi-Crop Leaf Disease Dataset of Tomato, Papaya, Zucchini, and Bottle Gourd from Bangladesh

Agri-Vision4 is an expert-validated multi-crop leaf disease dataset developed for computer vision and precision agriculture research, comprising 5,180 original images and 28,000 augmented images across 28 disease and healthy classes from four major crops: Bottle Gourd, Papaya, Tomato, and Zucchini. The original dataset is divided into 3,615 training images (70%), 762 testing images (15%), and 803 validation images (15%). To address class imbalance, augmentation techniques were applied to expand each of the 28 classes to 1,000 images, resulting in a balanced augmented dataset of 28,000 images. Images were collected under real-world agricultural conditions in Bangladesh using a SONY α7 II camera and standardized to 512×512 pixels. The augmented dataset provides a balanced representation across all disease and healthy classes, supporting robust model training and evaluation. Overall, Agri-Vision4 provides a diverse and balanced resource for plant disease classification, agricultural image analysis, computer vision, and deep learning–based precision agriculture research.