Computer vision, built ground-up
for food.
Most calorie trackers in 2026 are the same product they were in 2014, with a camera button bolted on top. PlateLens went the other way — computer vision is the core, and everything else (database, coach, fasting timer, league) is built around the assumption that you'll log by snapping a photo, not by typing.
That bet has paid off. PlateLens hit 4.8★ across App Store and Google Play, crossed 50,000+ active users, and trained its model on cuisines from over 100 countries — not just the US-supermarket food that dominates older databases. Pho, jollof rice, dal makhani, manakish, gimbap, koshari: the AI knows them, and it knows them by sight.
Honest about the limits. Photo portion estimates can be off ~10–20% on mixed dishes — casseroles, stews, layered bowls — where visual depth cues are ambiguous. We say this in our own marketing and we build the correction flow to take 2 taps. Restaurant low-light is harder than home overhead light; the app prompts for a better angle when it's struggling. And leagues + leaderboards aren't for everyone — every social feature is opt-out, with privacy-first defaults.
The trade-offs are deliberate. PlateLens isn't trying to be the deepest tracker (that's Cronometer) or the biggest database (that's MFP). It's the one that gets used every day, by people who don't want to file paperwork to count calories.