Detecting Disease in Ghana’s Maize plants


Maize accounts for over half of Ghana’s grain production, which reached historic highs in 2020. Yields have increased from 384,000 tonnes in 1971 to over 3 million tonnes in 2020, representing an average growth rate of 8.79% per annum.  

As global demand for maize continues to rise, it’s more important than ever to ensure that maize plants stay healthy and yields high. In Ghana and much of Africa, crop yield loss is a pressing issue that seriously impacts communities, food security, and growth at local and national levels.




Accra, Ghana

Company Size

50+ employees

Maize Disease in Africa

Maize in Africa faces threats from several diseases, but perhaps the most prolific in recent years is maize lethal necrosis (MLN), which affects maize in Africa as well as South America and South Asia. Once relatively confined to East Africa, MLN is beginning to take hold in other regions across the continent, including in Ghana. 

Another devastating set of diseases, maize streak disease (MSD), is one of the most pervasive crop threats in Sub-Saharan Africa, including Ghana, and accounts for around 10 to 15% of yield loss per annum. 

One of the most important aspects of maintaining maize yields is disease control, with preventable crop diseases regularly reaping devastation on farmers and communities. 

Supported a pro-bono initiative from Demeter Ghana, Aya Data looked to solve the problem of crop disease by creating a model to detect disease in maize plant leaves using computer vision.

Services: Polygon, Bounding Box, Semantic Segmentation

The Challenge

Demeter Ghana, a sustainability and agricultural services provider, wanted to end the misdiagnosis and improper treatment of common maize diseases in rural Ghana. 

The crux of the issue is that many of the common diseases affecting maize in the region look similar, leading to high misdiagnosis rates. In addition, farmers are often not sufficiently trained to recognize even severe infections, delaying swift and effective action to control their spread. 

Preventing year-on-year outbreaks is extremely difficult without an efficient and accurate way to identify and diagnose disease. Most diseases are treatable, providing they're found early, and treatment is less expensive than losing a significant percentage of crop yield. 

Traditionally, combatting maize disease relies on expensive and relatively scarce agronomists, who travel from farm to farm to dispense advice. Access to these agronomists is limited.  

Aya Data’s challenge was to create a crop disease identification model, accessible through an app, that would diagnose common maize diseases with a high degree of accuracy, and recommend treatment.

Our Solution

Aya Data used agronomic experts in Ghana to label 5000 images of diseased and non-diseased Maize plants. Not only were diseases bounded within the images, but tagged with classes denominating the disease type and severity.

Computer Vision Vs Crop Disease

The visual nature of crop disease is ideally suited to computer vision. Most maize diseases provide distinctive visual clues to the disease and its state of progression.

For example, the hallmark lesions of maize streak viruses (MSVs) are characteristic white vertical stripes that affect the leaves of maize plants. Northern corn leaf blight (NCLB) also features distinctive lesions, as does maize lethal necrosis disease (MLN), which causes the wilting and death of maize leaves. Other common fungal diseases cause varying speckled or spotted lesions. 

Our computer vision solution provides access to cutting-edge disease identification technology across the region. By self-identifying disease via an app, farmers can effectively reduce their reliance on expensive third-party agronomists, increasing their autonomy in the process. 

This fosters a culture of knowledge supported by technology - farmers can integrate such solutions into their traditional agricultural workflows without the need for on-site machinery or other interventions. The act of photographing diseases and receiving diagnosis and treatment information also helps educate agricultural workers in the region. 

Computer vision here is an intuitive and interactive tool, converting things we can see and feel into real-world insights and information. 

The Results

The images labelled by Aya Data were used to train a bespoke computer vision model to detect disease type and severity in maize plants from photos submitted by farmers, with 95% accuracy.


Assets Labelled

95% Accuracy

Achieved by Model


Of instances of Maize disease diagnosed

Explore Case Studies

Medical Imaging

Medical imaging is a time-intensive task that is being revolutionized by AI and ML.

Content Moderation For Advertising and Marketing

Reducing the risk of reputational damage for brands and businesses.

Ethical Data Sourcing in Africa

Get inspired by AyaData's case studies on data sourcing. See how we've helped businesses optimize their processes and achieve success. Learn more.

Image Segmentation in Satellite Imagery

Using satellite images to measure land-use changes in South America.

Real-Time Transcription of US Police Radios

Private companies are turning to technology to help combat crime and reduce the burden on society.

Identifying Shoplifting Events in Retail Stores

Companies are looking for innovative ways to curb their losses with the use of technology.

Preparing For The Growing Popularity of E-Scooters

The global electric scooters market size is expected to reach USD 34.7 billion by 2028.