PushtiPlate

PushtiPlate (পুষ্টি থালা) helps Bangladeshi mothers see what is missing from their baby's plate — with a single photo. In Bangladesh, only 29.5% of children aged 6–23 months receive a minimum acceptable diet (UNICEF / MICS 2025), 64.8% live in child food poverty, and 24% are stunted. Most mothers simply never learn which food groups their child is missing. A mother photographs her child's plate — no typing, no reading, no sign-up required. Muse Spark identifies the foods, and a deterministic scoring engine (transparent code, never AI guesswork) scores the plate against the WHO/UNICEF 8 food groups. The app suggests ONE affordable food to add (egg first), shows tomorrow's improved plate generated with Muse Image, and a Full Details section explains what is missing and what the next meal should be — in Bangla, English, or Hindi, with spoken Bangla audio for low-literacy users. Very low scores guide the mother to her community health worker. Built with Muse Spark, SAM 3.1, and Muse Image via the Meta Model API. Guest mode always works, and plate photos are stored privately in Supabase.