Project Details
Title:
Farm Doctor (Agriculture)
Details:
Smallholder farmers face devastating post-harvest losses of 30–50% from crop diseases they cannot properly diagnose or treat. The tools available to them fail on four fronts.
First, no connectivity: farms lack reliable internet, making cloud-dependent apps like Plantis unusable at the point of need. Farmers must travel to town just to attempt a diagnosis. Second, language barriers: recommendations arrive in English with technical jargon, while most target farmers speak Twi and have limited formal education. Third, inaccessible treatments: apps recommend imported agrochemicals with 2-week shipping times and $50 price tags, when local alternatives exist at GH₵50 from nearby shops. Fourth, incomprehensible instructions: "dilute 1:100 ratio" and "apply 2ml per litre" are meaningless without measuring equipment.
The result: farmers guess which disease they have, buy the wrong product, apply it incorrectly, and lose their harvest anyway. This cycle destroys livelihoods, reduces food security, and wastes the limited capital smallholders invest each season. The problem is not a lack of agricultural knowledge it is a last-mile delivery failure in getting the right knowledge to the right farmer in a form they can use.
Farm Doctor's primary objectives are: 1. Enable accurate, offline crop disease diagnosis using a hybrid AI approach — an on-device symptom matching engine for zero-connectivity environments, augmented by Claude Vision for photo-based diagnosis when connectivity is available. 2. Deliver treatment instructions in Twi using farmer-familiar measurements — replacing technical ratios with contextual analogies (bottle caps, weak tea colour, bucket quantities) that require no specialised equipment. 3. Connect farmers directly to affordable, local treatment sources by integrating a regional agro-dealer directory with WhatsApp ordering, ensuring every diagnosis leads to an actionable, affordable purchase. 4. Build a crowdsourced validation system where farmers report treatment outcomes by region, generating community-driven success rates that improve recommendation quality over time and build trust. 5. Demonstrate scalability from 4 crops and 5 regions to nationwide coverage across all 16 regions and 50+ diseases, with a clear integration pathway into the Complete Farmer network.
Ghana loses 30–50% of crop yield annually to plant diseases, disproportionately affecting smallholder farmers who grow chilli pepper, cassava, sweet potato, and groundnut. Existing diagnostic tools like Plantis fail in the Ghanaian context: they require constant internet (unavailable on most farms), deliver recommendations in English only (most smallholders speak Twi), prescribe imported agrochemicals costing $50+ (unaffordable), and use technical measurements like "dilute 1:100" that farmers cannot interpret. Farm Doctor responds to these Ghana-specific constraints directly. It works fully offline via PWA technology suited to 2G/3G coverage gaps. All diagnoses and treatment instructions are available in Twi. Measurements use familiar references "bottle cap of powder per bucket of water," "mix until weak tea colour." Treatment recommendations link to local agro-dealers via WhatsApp, with Ghana-market fungicides priced in cedis. Disease prevalence data is mapped across five Ghanaian regions (Ashanti, Greater Accra, Western, Volta, Northern), reflecting actual seasonal and climatic variation.
Source: