Making Citrus Health Information More Accessible
CitraScan is an open-source, no-cost educational citrus leaf screening prototype created by Ryan Haider, a high school student at Bella Vista High School in Fair Oaks, California.
The project began with the following question: Could a smartphone help a homeowner recognize when a citrus leaf might need closer attention?
Huanglongbing (HLB), also known as citrus greening, is a serious threat to citrus. While professional testing and expert evaluation are essential for determining disease status, homeowners may have difficulty recognizing concerning symptoms or knowing when to seek help. CitraScan is designed to serve as an educational first step.
How CitraScan Works
CitraScan analyzes an image of a citrus leaf and looks for visual patterns learned from its research datasets. The current prototype considers three broad categories: Healthy, HLB, and Nutrient deficiency
The model produces a prediction based on the visual patterns it detects. CitraScan also provides an image reliability score that considers characteristics such as lighting, blur, and image quality relative to the data used to train the model.
This is important because a model can have a confident prediction even when an input image is very different from the images it learned from.
Real-World Data Matters
One of the project’s current goals is to understand how well CitraScan performs outside research datasets.
Real citrus leaves can be photographed in many different conditions: direct sunlight, shade, different cameras, different backgrounds, damaged leaves, varying stages of symptoms, and many other situations.
This is why CitraScan is being developed with real-world feedback and community involvement in mind.
Growers, homeowners, gardening organizations, and citrus experts can help identify things such as:
- Conditions the current model struggles with
- Symptoms that are commonly confused with HLB
- Types of images missing from existing datasets
- Ways the interface could be more accessible
- Situations where a tool like CitraScan would or would not be useful
Long-Term Vision
The long-term goal is to develop an accessible educational resource for citrus health, particularly for people who may not have easy access to specialized agricultural equipment or expertise.
The project aims to remain:
Open-source. The technology and research can be examined and improved by others.
No-cost. The basic educational tool should not require specialized equipment or expensive software.
Hardware-agnostic. The goal is to make the system useful across a wide range of ordinary smartphones rather than designing it around a specific device category.
Community-informed. Experienced growers and agricultural experts should help direct how the project develops.
Help Improve CitraScan
CitraScan is still being developed, and community feedback is an important part of that process.
If you have citrus trees, experience with citrus health, or simply want to help test an educational technology project, you can try the demo and share your experience.
Try CitraScan: https://huggingface.co/spaces/RyanAI/CitraScan-v2
Give Feedback
GitHub (Work-In-Progress): https://github.com/Ryan-ProfProjects/CitraScan