TuinSnap is my own product, not yet released in the stores. The web version is already running. The app identifies plants, insects and animals from a photo and tracks what each plant needs and when.
It is built only for the Netherlands and Flanders and relies on existing sources: World Flora Online, Kew, GBIF, KNMI, PDOK and the Red Lists of the Netherlands and Flanders. An AI model proposes a scientific name, which is checked locally against the World Flora Online catalogue of about 1.7 million names. TuinSnap states no accuracy percentage.
Identification runs on Socket.io, so the result arrives in stages: first what is in the photo, then the details per species. Signing in uses a key pair per device.
One API serves three clients: the web app and the CMS in Next.js and the app in React Native. Underneath are Express and Prisma on MariaDB. The screens below come from a test build.
Screens
Features
- Identification of plants, insects and animals from a photo, with alternative candidates
- A care schedule per species that follows the season
- Watering advice based on KNMI weather data, with frost and heat warnings
- Garden drawing on PDOK aerial photos, with shade per hour
- Pet toxicity, with a pointer to NVIC or Antigifcentrum
Technologies
React Native
TypeScript
Next.js
Node.js
Express.js
MariaDB
Socket.io
Docker