AgroScan

Using machine learning models to detect diseases in sustitution crops.

Comment

What it does

AgroScan is an app that uses artificial intelligence and machine learning to help cocoa farmers identify diseases and pests in their crops. By taking a photo of the cacao plants and uploading it to the app, farmers can receive a real-time report on any potential issues. This can help farmers to quickly address problems and control the spread of diseases in their fields.

Inspiration

The idea for AgroScan came from the challenges faced by cacao farmers in Colombia, where cacao is often grown as a substitution crop in areas affected by armed conflict. This means that farmers in these areas may be unable to grow legal crops due to the violence and insecurity, so they turn to cacao to escape the danger. However, one of the major challenges facing these farmers is the prevalence of diseases in their crops, which can impact the sustainability of their operations and potentially lead to the loss of entire crops. AgroScan was developed to help these farmers identify and address diseases in their cocoa plants, improving their chances of success and contributing to the long-term sustainability of their operations.

How we built it

AgroScan was built using a combination of technologies, including next.js, tensorflow.js, and React hooks. The app features a convolutional neural network (CNN) trained to detect cacao crop diseases, which allows farmers to receive real-time reports on any potential issues with their crops. The app also includes features like image capture and submission of reports with GPS coordinates and farmer comments, as well as an admin panel that allows authorized users to view all submitted reports. The user interface was styled using tailwindcss, and the app is available in both English and Spanish to make it accessible to a wider audience.

Challenges we ran into

One of the major challenges we ran into while developing AgroScan was training the CNN to accurately detect cacao crop diseases. This required a significant amount of data and careful tuning of the model to achieve the desired level of performance and then make it performant to can embed it into a react component.

Accomplishments that we're proud of

  • Successfully collecting and curating a large dataset of images for training the CNN.

  • Building the app using JavaScript and tensorflow.js, allowing for efficient and effective training of the model.

  • Deploying the app publicly using vercel and railway, making it easily accessible.

  • Developing a CNN that is able to accurately detect cacao crop diseases in real-time, providing valuable assistance to farmers in identifying and addressing issues with their crops.

What we learned

Through the development of AgroScan, we learned about the challenges facing cacao farmers in Colombia and the importance of using technology to assist with sustainable agricultural practices. We also gained valuable experience in training and deploying machine learning models written in javascript.

What's next for AgroScan

In the future, we plan to continue improving and refining AgroScan to make it even more effective at helping cacao farmers identify and address diseases in their crops. We also plan to explore additional applications for the technology, such as identifying other types of crops or pests, in order to expand the scope of the app and make it even more useful to farmers.

Built With

Share this project:

Updates