GrassQuest: a local Gemma scavenger hunt that sends you outside

Description and use

GrassQuest is a local-first photo scavenger hunt app that uses Gemma 3 4B via Ollama, keeping all data on the device and operating offline.

Users want a privacy‑preserving, offline photo scavenger hunt that requires minimal screen time.

GrassQuest enables a quick, privacy‑preserving scavenger hunt that runs offline, with Gemma handling all AI tasks locally.

1. User enters destination, season, and time limit in the web UI.
2. Gemma 3 4B generates five items to find (e.g., leaf shape, bark, flower, puddle reflection, insect).
3. User takes photos of each item; the browser resizes to 1024px before upload.
4. Photos are sent to the Node.js/Express server, which calls Gemma to judge each photo and returns pass/fail.
5. After all items are judged, Gemma writes a short field journal using only the accepted photo notes.
6. The journal is displayed to the user.
7. Progress is saved in localStorage so a page refresh does not lose the quest.

It runs entirely on my laptop. No cloud, no account, and no photo ever leaves my machine.

I tested the photo check with real photos on my laptop. Five matching photos passed their tasks, and mismatched photos were rejected.

The model runs locally through Ollama, so a hosted copy would defeat the point.