X (Twitter) Space Event Discusses the Journey of the Winners of the Ghana Crop Disease Detection Challenge
The X (Twitter) Space discussion on “Developing Talents for the AI Ecosystem” provided valuable insights into how AI can revolutionise agriculture. The event featured the Ghana Crop Disease Detection Challenge winners, who shared their journey, challenges, and innovative solutions.
Mr. Elikplim Sabblah, Technical Advisor, FAIR Forward, GIZ, expressed his excitement about seeing Africans develop AI-driven solutions tailored to African problems. He hinted at how the Ghana Crop Disease Detection Challenge leveraged Afrocentric datasets collected by RAIL in collaboration with the Plant Protection and Regulatory Services Directorate (PPRSD) under Ghana’s Ministry of Food and Agriculture to develop a model to detect diseases on pepper, corn and tomato.
According to him, the competition fostered AI innovation and played a critical role in developing AI talent within the ecosystem. By addressing agricultural challenges using machine learning, participants demonstrated how AI could enhance disease detection, increase crop yields, and promote sustainable farming practices.
Ronny Polle (1st place), a Ghanaian medical doctor passionate about Machine Learning (ML) and AI research, mentioned the importance of understanding the problem before model development. His approach involved setting a threshold on the number of samples per class to improve accuracy and ensure the AI model was optimised for real-world agricultural applications.
Also, in the first-place position, Adeyinka Sotunde from Nigeria applied a YOLO-based object detection approach. He explained that YOLO (You Only Look Once) is a deep learning algorithm renowned for real-time object detection. He stated that YOLO’s mobile-friendliness and lightweight architecture make it ideal for farmers in remote areas to use offline. “The scalable model seamlessly integrates with farmers’ existing practices, enhancing traditional disease detection methods,” he added.
With 15 years of experience in healthcare-related data science in South Africa, Stefan Strydom (2nd place) focused more on model architecture than dataset refinement. He commended the competition’s clear engagement rules, which provided a solid framework for participants to innovate.
As a freelancer based in Nairobi, Raphael Laibuni Kiminya focused on improving dataset quality before training his model. He identified noise within the dataset and increased the number of correct samples. He then trained a baseline model to refine the dataset further, enhancing overall model accuracy.
In his closing remarks, Prof. Jerry John Kponyo, Principal Investigator and Scientific Director, RAIL, said the Ghana Crop Disease Detection Challenge is a unique platform for African innovators to apply AI in solving critical agricultural challenges while contributing to the broader AI ecosystem. He said RAIL remains committed to supporting AI talent development and fostering AI-driven innovations for sustainable agriculture and beyond.
