DigiKam Face Recognition Database Reader

A C# console application that extracts face tagging data from DigiKam databases for use in facial recognition systems, machine learning pipelines, and photo organization workflows.

Overview

DigiKam is a professional photo management application that includes advanced face recognition capabilities. This tool extracts and consolidates face tagging data from DigiKam's SQLite databases, making it easy to:

• Build ML Training Datasets: Export tagged photos for training custom facial recognition models
• Organize Photo Collections: Quickly locate all photos of specific individuals
• Analyze Face Tagging: Understand your photo collection's face recognition coverage
• Integrate with Pipelines: Generate file lists for batch processing in ML workflows

Features

• Database Statistics: View comprehensive metrics about identities, tags, and images
• Person Discovery: List all recognized people in your photo collection
• Smart Search: Find people by name with fuzzy/partial matching
• Image Retrieval: Locate all photos containing specific individuals with full metadata
• Export Functionality: Generate text files of image paths for downstream processing
• Safe Read-Only Access: Opens databases in read-only mode to prevent accidental modifications
• Multi-Database Support: Consolidates data from multiple DigiKam databases:
  • recognition.db - Face recognition data and person identities
  • digikam4.db - Main image database with tags and metadata
  • thumbnails-digikam.db - Thumbnail storage

Requirements

• .NET 6.0 SDK or later
• DigiKam photo management software (to generate the databases)
• DigiKam database files in accessible location
• Windows, Linux, or macOS

Installation

1. Clone or download this repository
2. Navigate to the project directory:

   cd facerecog_Db_reader

3. Build the project:

   dotnet build

Configuration

The application looks for DigiKam databases in the following default location:

• Windows: %USERPROFILE%\Pictures\DH
• Linux/macOS: $HOME/Pictures/DH

If your DigiKam databases are in a different location, update the path in Program.cs:8.

Database Structure

DigiKam stores data across multiple SQLite databases:

recognition.db

• Identities: Person IDs and types
• IdentityAttributes: Person names and attributes (join on id where attribute='name')
• FaceMatrices: Face embeddings/encodings linked to identities

digikam4.db

• Images: Image file information and metadata
• Albums: Album/folder hierarchical structure
• AlbumRoots: Base paths for album collections
• Tags: All tags including person names created by face recognition
• ImageTags: Many-to-many relationship linking images to tags
• ImageInformation: Extended metadata (creation dates, ratings, etc.)

Usage

1. Build the project:

   cd facerecog_Db_reader
   dotnet build

2. Run the application:

   dotnet run

3. Menu options:

   • 1. List all people: Display all unique person names from the database
   • 2. Search for a person: Search for people by name pattern
   • 3. Find images by person name: Get all images containing a specific person
   • 4. Exit: Close the application

Example Workflow

1. Start the application - it will show database statistics
2. Choose option 2 to search for a person (e.g., enter "John")
3. Once you find the exact name, use option 3 to find all images
4. Export the image paths to a text file for further processing

Project Structure

facerecog_Db_reader/
├── Models/
│   ├── Identity.cs          # Person identity model
│   ├── IdentityAttribute.cs # Person attributes
│   ├── Image.cs             # Image metadata
│   └── Tag.cs               # Tag information
├── FaceRecognitionDbReader.cs  # Main database reader service
├── Program.cs               # Console interface
└── TODO.md                  # Project roadmap

Implementation Phases

Phase 1: Read the Database ✓

• Implemented SQLite database reading
• Created data models for all entities
• Added database schema exploration

Phase 2: Analyze the Data ✓

• Statistics collection
• Person search functionality
• Data aggregation across multiple databases

End Goal: Search Images by Person ✓

• Search for people by name
• Retrieve all images for a person
• Export image paths to files

Notes

• The application opens databases in read-only mode for safety
• Image paths are constructed from AlbumRoots, Albums, and image names
• Names can come from either the recognition.db Identities or digikam4.db Tags