Executive summary
This project aims to develop an artificial intelligence-based mobile application capable of instantly detecting rice diseases from photos taken by farmers — without requiring an internet connection. The app will provide immediate diagnosis and tailored treatment recommendations. The pilot will be conducted in Madagascar, a country where rice is vital to food security. By leveraging computer vision and offline AI inference, this project directly supports ACFAI’s mission to harness AI for sustainable development and poverty reduction across Africa.
Background and rationale
Rice farming, covering approximately 1.5 million hectares in Madagascar, is a cornerstone of national food security (Rakotomalala et al., 2019). However, it is seriously threatened by multiple plant diseases, particularly Rice Yellow Mottle Virus (RYMV) and bacterial diseases like Bacterial Leaf Blight (BLB) and Bacterial Leaf Streak (BLS).
RYMV, recently introduced from East Africa, has rapidly spread throughout Madagascar in just a few decades. It can cause severe yield losses, and the Malagasy strain (S4-Mg) is genetically distinct (Rakotomalala et al., 2019). BLB, first officially reported in Madagascar in 2019, can lead to yield losses exceeding 70% under severe conditions (Raveloson et al., 2023), while BLS can cause 15–20% losses depending on climate and rice variety (FOFIFA, 2020).
In rural areas, farmers lack access to diagnostic tools and reliable internet connections, resulting in delayed or incorrect disease management. A mobile application using offline AI image recognition offers a practical, scalable, and inclusive solution to mitigate these challenges.
Objectives
– A fully functional, offline AI mobile app capable of diagnosing at least 3 key rice diseases with >90% accuracy.
– 200+ farmers trained and actively using the app in the pilot phase.
– 20–40% average yield improvement reported among participating users due to earlier and more accurate intervention.
– Creation of a rice disease image repository for Madagascar (~10,000+ labeled images).
– A documented model for scalable deployment in other African rice-producing countries.
Expected results
– A fully functional, offline AI mobile app capable of diagnosing at least 3 key rice diseases with >90% accuracy.
– 200+ farmers trained and actively using the app in the pilot phase.
– 20–40% average yield improvement reported among participating users due to earlier and more accurate intervention.
– Creation of a rice disease image repository for Madagascar (~10,000+ labeled images).
– A documented model for scalable deployment in other African rice-producing countries.
Impacts
– Direct benefits: Improved yields, income stabilization, and reduced pesticide misuse for smallholder farmers.
– Digital inclusion: Empowerment of underserved rural communities through access to AI-based diagnostics.
– Scientific contribution: Advancement in localized, offline AI tools for precision agriculture in Africa.
– Policy influence: Data generated can inform national strategies on plant health and digital agriculture.
References
- FOFIFA. (2020). Les maladies bactériennes du riz à Madagascar [Poster]. Antsirabe : dP SPAD – DINAMICC.
- Harinjaka, R. (2020, septembre). Les maladies bactériennes du riz : Poster BLB – BLS [Poster]. FOFIFA. https://www.fofifa.mg/wp-content/uploads/2020/09/PosterBLBBLS‑VF‑Final‑.pdf
- Hughes, D. P., & Salathé, M. (2019). An open access repository of images on plant health to enable machine learning applications. Virus Evolution, 5(2), vez023. https://doi.org/10.1093/ve/vez023
- Poulin, L., Raveloson, H., Sester, M., Raboin, L.-M., et al. (2014). Confirmation of bacterial leaf streak caused by Xanthomonas oryzae pv. oryzicola on rice in Madagascar. Plant Disease, 98(10), 1423. https://doi.org/10.1094/PDIS-02-14-0132-PDN
- Rakotomalala, M., Vrancken, B., Pinel-Galzi, A., Ramavovololona, P., Hébrard, E., Randrianangaly, J. S., … & Fargette, D. (2019). Comparing patterns and scales of plant virus phylogeography: Rice yellow mottle virus in Madagascar and in continental Africa. Virus Evolution, 5(2), vez023. https://doi.org/10.1093/ve/vez023
- Rakotomanana, H. M. (2023). Phytopathologie : Importance des maladies du riz à Madagascar et stratégies de lutte [Mémoire de fin d’études, Ecole Supérieure des Sciences Agronomiques, Université d’Antananarivo].
- Raveloson, H., Rabekijana, R., Rakotonanahary, N. M., Szurek, B., Muller, B., vom Brocke, K., & Hutin, M. (2023). First report of bacterial leaf blight disease of rice caused by Xanthomonas oryzae pv. oryzae in Madagascar. Plant Disease, 107(8), 2510. https://doi.org/10.1094/PDIS-03-23-0411-PDN