#2222nd of 22 in ML frameworks and libraries
face_recognition
Python library and CLI for face detection, landmarks and matching, built on dlib
Live demo ↗
(opens in a new tab)Documentation ↗
(opens in a new tab)Repository on GitHub ↗
(opens in a new tab)
- Stars
- 56.8k
- License
- MIT
- Last commit
- Jun 2026
- Last release
- Apr 2018
- Language
- Python
55,850 on 25 Mar 202656,821 on 14 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Python library built on dlib that finds faces in images, returns landmarks for eyes, nose, mouth and chin, and compares face encodings to identify people. It also installs `face_recognition` and `face_detection` CLI tools that process folders of images, with a `--tolerance` setting (default 0.6) and `--cpus` for parallel runs. The README cites 99.38% accuracy on Labeled Faces in the Wild.
Who it is for: Python developers adding face detection or matching to scripts and apps
Strengths
- Few-line Python API: face_locations, face_landmarks, face_encodings, compare_faces
- CLI tools identify faces across image folders, with --cpus for multi-core runs
- Many runnable examples: webcam, video file, Raspberry Pi, Flask service, KNN and SVM
- Optional CNN detector with GPU batch support when dlib is built with CUDA
Weaknesses
- Last release is 2018-04-02, so PyPI lags the repository
- Requires building dlib with cmake; Windows is not officially supported
- README says accuracy is weaker on children and varies between ethnic groups
- CNN detection model needs a GPU for good performance
What it needs
- GPU optional
- Docker + Compose
- Needs dlib, cmake, Python 3.3+ or 2.7, OpenCV (webcam/video examples), Flask (web service example)
- Models: dlib HOG face detector, dlib CNN face detector, dlib ResNet face encodings
Also in ML frameworks and libraries
See all 22| Rank | Project | Score |
|---|---|---|
| 1 | transformersModel-definition library for text, vision, audio and multimodal models | 94 out of 100 |
| 2 | pytorchPython tensor library with GPU support and tape-based autograd for deep learning | 93 out of 100 |
| 3 | ultralyticsPython library and CLI for training and running YOLO vision models | 88 out of 100 |
| #4 | supervisionPython toolkit for loading, annotating and counting computer vision detections | 84 out of 100 |
| #5 | pytorch-image-modelsPyTorch library of image models, pretrained weights and reference training scripts | 82 out of 100 |