Donald O. Ebuga

In Sub-Saharan Africa, smallholder farmers produce over 80% of staple food crops (cassava, maize, yam, tomato, sorghum), yet lose up to 40% of their harvest annually to destructive foliar diseases and pests like Cassava Mosaic, Southern Corn Rust, and Fall Armyworm. Most existing AI agricultural apps fail in the field due to two systemic barriers: 1. The Connectivity Divide: Rural farms operate with expensive, spotty, or zero cellular data. Cloud-only tools fail where crops actually grow. 2. The Language & Literacy Divide: Conventional agronomy manuals provide dense, academic English text, alienating low-literacy farmers who speak indigenous languages and everyday dialects. AgriPulse Edge is an offline-first crop intelligence system and multilingual voice copilot designed to bridge this last mile. Using a single leaf photograph taken with any basic smartphone, AgriPulse delivers: - Instant Pathology Identification: Recognizes diseases across major African staple crops with visual leaf validation that rejects non-leaf images to prevent hallucinations. - SAM-Powered Lesion Quantification: Utilizes Meta’s Segment Anything Model (SAM) to isolate diseased lesions from healthy leaf tissue and calculate the exact percentage of damage (e.g., 24% moderate vs. 68% severe). - Culturally Grounded Organic Prescriptions: Replaces unaffordable commercial pesticides with 3 actionable, zero-cost remedies prepared from locally accessible materials (fresh neem/dogonyaro leaf steeps, sieved wood ash, and local black soap stickers). - 5-Language Spoken Voice Notes: Synthesizes audible voice guidance in English, Nigerian Pidgin, Hausa, Yoruba, and Igbo with authentic prosody, cadence, and tone-mark normalization for illiterate farmers. AgriPulse delivers this intelligence across three unified tiers: 1. Rural Farmer: Direct conversational WhatsApp bot via Meta WhatsApp Business Cloud API—zero app installation required, supporting voice notes, photos, and local dialects. 2. Field Extension Agent: Offline-first mobile PWA and React Native app with an offline queue (AsyncStorage) that captures field photos with zero data and synchronizes idempotently upon reconnection. 3. Cooperative Manager: Web command workspace at agripulse.zentreeca.org/app featuring lesion opacity inspection, disease outbreak records, and treatment verification slips. Web Project Link: https://agripulse.zentreeca.org/