Case study

Green Illness Detection

We're reinventing turfgrass monitoring through advanced image analysis

Client: Mint S.r.l.

The Project

Green Illness Detection is an innovative project that leverages computer vision and artificial intelligence for the monitoring and early identification of diseases in turfgrass, with a particular focus on golf courses.

Developed in collaboration with Mint S.r.l., a Rovereto-based company specializing in IoT solutions for precision agriculture, the project aims to revolutionize the management and maintenance of green areas.

The project is part of the "Foundation Open Factory" initiative, promoted by ELIS and funded by Trentino Sviluppo.

Challenges

The Challenges Ahead

Maintenance Costs

Late detection of turfgrass diseases leads to broader and more expensive maintenance work, with consequent increases in labor and materials needed.

Environmental Impact

The more the disease spreads, the more water and pesticides are needed to treat it, with a significant impact on the ecosystem and a risk of overuse of chemicals.

The lack of an effective real-time monitoring system makes it difficult to detect turfgrass diseases promptly, leading to late interventions that result in higher economic and environmental costs.

Goals

Project Goals

Reduce Costs

Minimize maintenance expenses through early disease identification, enabling targeted interventions that require fewer resources and less labor.

Optimize Resources

Plan labor more efficiently, allowing staff to focus only on areas that actually need intervention.

Reduce Environmental Impact

Limit the use of water and pesticides through early targeted interventions, contributing to the environmental sustainability of golf courses.

AI solution

Our Solution

We developed a computer vision algorithm capable of identifying, recognizing and precisely localizing turfgrass diseases in their very earliest stages.

Capture

Cameras mounted on maintenance machinery capture turfgrass images during routine operations

Analysis

Computer vision algorithm that analyzes the images to identify and localize turfgrass diseases

Visualization

Interactive course map that shows at-risk areas with their GPS coordinates for targeted interventions

The system integrates image analysis data with geolocation, creating a health map of the golf course that enables precise and timely interventions.

Roadmap

Implementation Phases

1. Preliminary Study

Analysis of the scientific literature and existing public datasets on plant disease detection, to understand the state of the art in computer vision applied to agriculture.

2. Image Collection

In collaboration with Mint, we conducted on-site visits at golf courses, capturing numerous images of turfgrass exhibiting various types of diseases at different development stages.

3. Image Labeling

The collected images were manually labeled, precisely indicating regions where diseases were present, creating a supervised dataset for model training.

4. Initial Training

We selected the most suitable neural network architecture and trained the model using the created dataset, optimizing parameters to maximize recognition accuracy.

5. Iteration & Optimization

As new data was collected, we continuously improved the model, increasing its precision and ability to recognize different diseases under varying conditions.

6. Validation & Integration

We tested the system on a sports facility in Trentino, verifying recognition accuracy and integrating it with Mint's IoT platform to provide complete real-time monitoring.

Results

Concrete Benefits

The system was validated on a Trentino sports facility, demonstratinghigh accuracy in detecting different turfgrass diseases.

Thanks to early disease identification, it's possible to intervene promptly, significantly reducing both maintenance costs and environmental impact.

The results show that targeted interventions based on data provided by our system require fewer resources, labor, and chemicals, contributing to more sustainable and economical management of golf courses.

Reduced Maintenance Costs

Early disease identification allows more targeted and less extensive interventions, with significant savings in resources and labor. Course administrators can better plan maintenance operations, focusing efforts only where really needed.

Reduced Environmental Impact

More efficient use of water and pesticides helps limit the environmental impact of maintenance operations. Lower chemical use makes the course healthier for users and the surrounding ecosystem, promoting more sustainable management of green areas.

Thanks to AI algorithms based on video stream analysis, it's possible to monitor green areas in real time, automatically detecting potential diseases and providing effective tools for optimal course management.

Case studies