Ansikten

Face recognition tool for event photography.

---

What is this?

Ansikten helps photographers identify and rename people in large collections of RAW images (NEF). It consists of:

• Backend - Python CLI for batch processing + FastAPI server
• Frontend - Electron workspace for interactive review

The tool is designed for sports events, school activities, or any context where many people appear across multiple photos.

---

Quick Start

Backend

cd backend
pip install -r requirements.txt

# Process images
./hitta_ansikten.py 2024*.NEF

# Rename based on detected faces
./hitta_ansikten.py --rename --processed .

Frontend

cd frontend
npm install
npm run build:workspace
npx electron .

Backend auto-starts on http://127.0.0.1:5001

---

Documentation

For Users

• Installation - Download prebuilt packages
• Getting Started - Development setup
• CLI Reference - Command line usage
• Workspace Guide - GUI usage
• Keyboard Shortcuts - All shortcuts

For Developers

• Architecture - System overview
• API Reference - REST and WebSocket API
• Database - Data files and formats
• Release Guide - Publishing releases
• Building - Build from source
• Theming - CSS variable system
• Contributing - Git workflow and code style
• Onboarding - New developer guide

---

Features

• Batch face detection in RAW files (NEF)
• Face recognition against known person database
• Interactive review - confirm, reject, or manually name faces
• Automatic file renaming based on detected people
• InsightFace face recognition (512-dim embeddings, cosine distance)
• Modular workspace - GIMP-like dockable panel UI
• Real-time updates via WebSocket

---

Project Structure

ansikten/
├── backend/          # Python CLI + FastAPI server
├── frontend/         # Electron workspace (FlexLayout)
├── shared/           # Common type definitions
└── docs/             # Documentation
    ├── user/         # User guides
    └── dev/          # Developer guides

---

Data Storage

All persistent data in ~/.local/share/faceid/:

File	Description
encodings.pkl	Known faces database
processed_files.jsonl	Files already processed
attempt_stats.jsonl	Processing attempt log
config.json	User configuration

---

Technology

Backend:

• Python 3.9+
• InsightFace (ONNX Runtime)
• FastAPI, WebSocket, rawpy

Frontend:

• Electron
• React + FlexLayout
• Canvas-based image rendering

---

License

GPL-3.0