cmusatyalab/openface

What it solves

OpenFace is a free and open-source library designed for face recognition using deep neural networks, providing a general-purpose framework for identifying people from images or video streams.

How it works

The library uses deep neural networks to generate representations (embeddings) of faces. It includes tools to process batches of images, compare two images for similarity, and train custom classifiers. It can be integrated into web applications or used with webcam streams for real-time identification.

Who it’s for

It is primarily intended for researchers and developers who need a face recognition system that can be run on mobile applications or general-purpose computing environments.

Highlights

• Real-time capabilities: Includes demos for web-based and webcam-stream face recognition.
• Batch processing: Tools to generate face representations from large sets of images.
• Custom training: Scripts provided to train new neural network models and classifiers.
• Evaluation tools: Includes scripts to evaluate accuracy using the LFW (Labeled Faces in the Wild) dataset.