Quickstart Guide¶
Get up and running with CUVIS.AI in 5 minutes.
Installation¶
First, ensure you have Python 3.10+ and uv installed:
# Clone the repository
git clone https://github.com/cubert-hyperspectral/cuvis-ai.git
cd cuvis-ai
# Install dependencies
uv sync
See the Installation Guide for detailed setup instructions.
Download Sample Data¶
Download the Lentils dataset from Hugging Face:
# Automated download (default: lentils dataset)
uv run download-data
# Or explicitly specify dataset
uv run download-data --dataset lentils
This downloads ~1.0 GB of real hyperspectral data to data/Lentils/.
Quick Demo: Run Pre-Trained Pipeline¶
Want to see CUVIS.AI in action first? Run inference with a pre-configured pipeline:
# View pipeline structure
uv run restore-pipeline --pipeline-path configs/pipeline/rx_statistical.yaml
# Run inference on sample data
uv run restore-pipeline --pipeline-path configs/pipeline/rx_statistical.yaml --cu3s-file-path data/Lentils/Demo_000.cu3s
This loads the pipeline configuration and runs anomaly detection on the sample hyperspectral cube.
Train Your Own Pipeline¶
Train an RX anomaly detector from scratch:
Results are saved to outputs/base_trainrun/.
What Just Happened?¶
- Loaded data - The Lentils hyperspectral dataset
- Built pipeline - RX statistical anomaly detector from
configs/pipeline/rx_statistical.yaml - Trained model - Statistical initialization on training data
- Saved results - Pipeline, weights, and metrics to
outputs/
Use Your Trained Model¶
After training, restore and use your model for inference:
# Restore trained pipeline
uv run restore-pipeline --pipeline-path outputs/base_trainrun/trained_models/RX_Statistical.yaml --cu3s-file-path data/Lentils/Lentils_000.cu3s
The pipeline will load your trained weights and run inference on new data.
Next Steps¶
Learn the fundamentals:
- Core Concepts Overview - Understand the architecture
- Configuration Basics - Master Hydra composition
Follow comprehensive tutorials:
- RX Statistical Tutorial - Statistical anomaly detection
- Channel Selector Tutorial - Learnable band selection
- Deep SVDD Tutorial - Deep learning approach
Explore how-to guides: