Search and identify contacts by physical traits, gathered from pictures with AI.

Getting Started

Requirements

• Docker/Docker Compose
• make command

1. Create a local .env file

make prepare-env

2. Build maven project and container images

With GPU support (for NVIDIA only):

• Install the NVIDIA Container Toolkit.

make build-gpu

Without GPU support:

make build

3. Run containers

make up

Ollama models

This project uses a vision-capable (multimodal) model. You can explore all available options here to find one that best fits your use case.

By default, the project uses gemma3:12b-it-qat. To change it, just update OLLAMA_MODEL in your .env or shell.

Make sure to pick a model size that your GPU/CPU and memory can comfortably support.

Demo

You can try the project online at: https://eaglai.griffin-frog.ts.net/.

• Database refresh: The database is automatically refreshed every hour. All data will be reset at that time.
• Mock Ollama responses: This demo uses the mock configuration, so the Ollama API responses are static and not representative of any actual parsed facial features.

⚠️ Warning: Do not enter any personal or sensitive information. All data is temporary and publicly accessible.

License

This project is licensed under the MIT License.