Colombian coffee tree leaves multispectral images dataset.

Area of Science:

• Agricultural Science
• Computer Vision
• Remote Sensing

Background:

• Coffee rust (Hemileia vastatrix) significantly impacts global coffee production.
• Accurate and early disease detection is crucial for effective crop management and yield optimization.
• Existing methods for disease identification may lack the detailed spectral information needed for early-stage detection.

Purpose of the Study:

• To present a comprehensive, high-quality multispectral image database of coffee leaves.
• To facilitate research in precision agriculture for disease detection using advanced image analysis.
• To enable the development of machine learning models for identifying coffee rust.

Main Methods:

• Collected and curated a dataset of 6726 multispectral images of coffee leaves.
• Captured images in JPG (RGB) and TIF (16-bit, 5 multispectral bands: blue, green, red, red-edge, NIR) formats.
• Ensured image quality and consistency through controlled lighting conditions and standardized labeling.

Main Results:

• A unique, large-scale database of coffee leaf images across multiple spectral bands is now available.
• The database includes images of both healthy leaves and leaves exhibiting lesions from coffee rust.
• Spectral data ranges from blue (450-500 nm) to near-infrared (750-900 nm), with a specific red-edge band (approx. 840 nm).

Conclusions:

• The presented multispectral image database is a valuable resource for agricultural research.
• It supports the development of advanced image processing and machine learning techniques for crop disease detection.
• This resource can lead to improved early detection and management of coffee rust, enhancing farm productivity.