TerraQuest AI — Your Offline AI Companion for the Real World

Local AI & Privacy Controls

Settings & Local Inference Architecture 🌿

Configure your local open-source models, test Ollama connections, adjust your privacy profile, and manage offline data.

Select your local inference engine and open-weight model.

Local-First

Inference Intelligence Provider

Ollama Endpoint URL

Open-Weight Model

Use internal API proxy (recommended to bypass browser CORS)

Species Vision Intelligence: Instant Model or Kaggle Training

Open-Source AI

To classify species with high accuracy and confidence upon upload, you have two open-source options:

1. Instant Vision Model (Ollama)

Zero training needed. Pull lightweight open weights with zero cloud telemetry:

ollama pull moondream
# Or for 7B/11B GPUs:
ollama pull llava

2. Train with Kaggle Datasets

Use the bundled pipeline to fine-tune MobileNetV3 / EfficientNet on Oxford Flowers or Birds 525:

python training/download_dataset.py --dataset flowers102
python training/train_classifier.py --dataset flowers102

See README.md (Section 7: Machine Learning & Kaggle Suite) for step-by-step dataset preparation, ONNX export, and evaluation metrics.

Preferences used exclusively to tailor outdoor suggestions. Stored locally.

Explorer Name

Experience Level

Screen-Free Goal (Mins / Week)

Full control over your device storage. Zero external telemetry.

Export Full Field Notebook Backup

Downloads observations, missions, garden calendars, and challenges as JSON.

Reset Application State

Wipes IndexedDB and restores factory seed data.