Presenters:
Vaishnavi Sen
Serineh Megerdichian
Denise Pacheco
California State University, Northridge
Cocoa farming, a cornerstone of Colombia’s agricultural economy and a vital source of
livelihood for rural farmers, is critically threatened by plant diseases such as black pod rot,
monilia, and pod borer. These diseases reduce yield and quality, causing a 30-50% revenue
loss for farming families.
This research introduces a geospatially informed approach to diseased crop management,
integrating machine learning and environmental data for precision agriculture. A deep learning framework, including convolutional neural networks (CNNs), processes visual data of cocoa pods to identify disease symptoms with high accuracy. Complementing this is a network of hardware sensors and GPS modules that collect critical environmental metrics, including temperature, humidity, and elevation.The project incorporates data from multiple farms to enhance model robustness and scalability.
By integrating information from diverse regions, the model can better account for geographic and environmental variations in disease patterns. This approach improves the system’s accuracy and fosters collaboration among farming communities, promoting wider adoption of sustainable practices.
By combining the image and geospatial data, the model could provide location-specific disease predictions and identification of disease hotspots, fostering sustainable farming practices. This interdisciplinary effort leverages image-based diagnostics alongside geospatial analytics to deliver an innovative, scalable solution for enhancing cocoa yields and supporting Columbia’s rural farming communities.
