ageitgey/face_recognition

Project overview

A Python library and command-line tool for detecting faces, finding facial landmarks, and identifying people in images using dlib-based deep learning models. It achieved 99.38% accuracy on the Labeled Faces in the Wild benchmark but requires medium setup effort and is not officially supported on Windows.

Project type

Image & Vision

Use cases

Image Processing

Deployment

Refer to project documentation

License

MIT

Best for

• Developers seeking a Python API or command-line interface to detect adult faces, extract facial landmarks, or match faces against known encodings on self-hosted systems.

Key capabilities

• Locates all faces in a picture and returns their pixel coordinates. Optionally supports a more accurate deep-learning-based face detection model (CNN).
• Finds the locations and outlines of each person's eyes, nose, mouth, and chin.
• Recognizes who appears in a photo by comparing face encodings against known images.
• Provides `face_recognition` to recognize faces in photographs or folders and `face_detection` to find face locations.
• Manipulates facial features for digital make-up application.
• Speeds up face recognition by processing images in parallel using multiple CPU cores.

Limitations and risks

• Windows is not officially supported.
• The face recognition model is trained on adults and does not work very well on children.
• Accuracy may vary between ethnic groups.
• A bug in the CUDA libraries on the Nvidia Jetson Nano will cause this library to fail silently if specific instructions are not followed.

Getting started

• Setup difficulty is medium. First, install dlib from source with Python bindings, then install cmake, and finally run pip3 install face_recognition.

Evidence and sources

• README: Recognize and manipulate faces from Python or from the command line with the world's simplest face recognition library.
• GitHub project description: The world's simplest facial recognition api for Python and the command line
• README: The model has an accuracy of 99.38% on the Labeled Faces in the Wild benchmark
• Release: v1.2.2
• README: Python 3.3+ or Python 2.7 * macOS or Linux (Windows not officially supported, but might work)

Reviewed at: 2026-07-17T06:28:47Z