Foliage Hunter: a Colorado leaf guide on a fine-tuned Gemma 3, built to keep the screen short

Description and use

Foliage Hunter lets users quickly identify fall foliage in Colorado using a fine‑tuned Gemma 3 model and Magpie TTS, keeping the walk uninterrupted.

Problem

Users on the Front Range want to identify fall foliage quickly without stopping to read.

Outcome

The app enabled a ten‑minute field test in Castle Rock, Colorado, successfully identifying six species and recording geocoded finds. The initial HEIC‑to‑JPEG conversion fix reduced payloads and resolved 500 errors.

Workflow

• 1. Walk and find a leaf
• 2. Tap to capture photo
• 3. Model identifies the tree and generates a short spoken description
• 4. TTS speaks the description while the photo is processed
• 5. The find is pinned to a trail view showing all previous finds

Evidence

Foliage Hunter is a pocket leaf guide for peeping fantastic fall foliage. You walk, you find a leaf, you snap it.

Source

The loop is deliberately short. One tap to shoot, a few seconds of a leaf wreath spinning, and the guide starts talking before the whole reply has even been synthesized.

Source

The fix was to re‑encode on the phone with expo-image-manipulator (resize to 1280 px, JPEG at 0.82) before upload, which also cut the payload to a fraction of the original.

Source

The app runs in Expo Go on your phone and talks to a model you host.

Source