A Multi-Class Guava (Psidium guajava L.) Leaf Diseases Dataset for Precision Agriculture and AI-Based Crop Health Monitoring.

This dataset presents a comprehensive collection of guava (Psidium guajava) leaf images systematically collected from orchards in Savar and Jatrabari districts under natural field conditions of Bangladesh, encompassing multiple disease conditions and physiological states across different diseases. The dataset contains 3,355 images categorized into six classes: Anthracnose, Dried, Healthy, Sooty_Mould, Senescence, and Insect Infestation. All images were captured diversity from November 2025 to January 2026 using smartphone cameras in real agricultural environments, incorporating natural variations in illumination, background complexity, leaf orientation to ensure realistic. This dataset serves as a valuable benchmark for machine learning and deep learning research, supporting applications in automated disease detection, crop health monitoring, plant pathology analysis, and precision agriculture, ultimately contributing to sustainable guava production and improved crop management strategies.

Captured Using:
Two Android smartphones, Google Pixel 7 and Samsung Galaxy A56 5G

Dataset Classes and Image Counts:
Anthracnose: 773 images
Dried: 271 images
Healthy: 831 images
Insect_Infestation: 441 images
Senescence: 510 images
Sooty_Mould: 529 images

Total images: 3,355

Image Details:
Resized image resolution: 512 x 512
Image formate: JPG
Color mode: RGB
Location: Savar, Jatrabari, Bangladesh.