davisking

license

BSL-1.0(Boost Software License 1.0, OSI-approved

source

public(github.com/davisking/dlib)

core features withheld

no — maintained by its author

model-zoo-terms

public-domain(dlib-models released into the public domain, except shape_predictor_68_face_landmarks, whose ibug 300-W training data excludes commercial use)

dlib is under the Boost Software License 1.0, a short permissive license the OSI approves, and builds from the public repository its author maintains; nothing is sold. Its trained models are published separately and downloaded by hand, mostly in the public domain. The 68-point landmark model is the exception: its training data excludes commercial use. The face-recognition pipeline dlib is credited for uses the public-domain 5-point landmark model and recognition network instead, so the restricted model does not govern. BSL-1.0 is OSI-approved, so the license tier is osi.

• https://pypi.org/pypi/dlib/json recorded 2026-09-27

  PyPI record for dlib 20.0.1, uploaded 2026-03-29, license "Boost Software License", Homepage github.com/davisking/dlib.

• https://raw.githubusercontent.com/davisking/dlib-models/HEAD/README.md recorded 2026-09-27

  dlib-models README: "anyone can do whatever they want with these model files as I've released them into the public domain"; for the 68-point landmark model, "The license for this dataset excludes commercial use".

• https://raw.githubusercontent.com/davisking/dlib/HEAD/LICENSE.txt recorded 2026-09-27

  LICENSE.txt is the "Boost Software License - Version 1.0 - August 17th, 2003", "Permission is hereby granted, free of charge, to any person or organization".

• https://raw.githubusercontent.com/davisking/dlib/HEAD/python_examples/face_recognition.py recorded 2026-09-27

  Face-recognition example: "./face_recognition.py shape_predictor_5_face_landmarks.dat dlib_face_recognition_resnet_model_v1.dat ../examples/faces", with both models fetched by hand from dlib.net/files.

• https://raw.githubusercontent.com/davisking/dlib/HEAD/README.md recorded 2026-09-27

  README: "Dlib is a modern C++ toolkit containing machine learning algorithms and tools for creating complex software in C++ to solve real world problems."; no paid tier or commercial edition mentioned.

• https://ungh.cc/repos/davisking/dlib/files/master recorded 2026-09-27

  Repository tree of 2,304 paths with no ee, enterprise, commercial or pro directory.

Measured on monthly PyPI downloads of the dlib package, which ships only as source and compiles on install. The C++ library used directly is not counted.

• https://pypistats.org/api/packages/dlib/recent recorded 2026-09-27

  last_month 174,853 downloads of dlib

dlib is first a general machine-learning toolkit with many learners to choose among, but it also ships a trained face-recognition network that compares faces the user enrolls without any training run, the same capability that places InsightFace here. Outside faces, choosing and training a model for a new task is left to the user.

• http://dlib.net/ml.html recorded 2026-09-27

  Machine Learning page "Primary Algorithms": Binary Classification (svm_c_trainer, rvm_trainer, svm_pegasos, auto_train_rbf_classifier ...), Multiclass Classification, Regression (krr_trainer, rvm_regression_trainer, random_forest_regression_trainer, svr_trainer ...), Structured Prediction, Clustering (kkmeans, spectral_cluster, chinese_whispers ...), and Deep Learning layers.

• https://raw.githubusercontent.com/davisking/dlib-models/HEAD/README.md recorded 2026-09-27

  dlib-models README: "dlib_face_recognition_resnet_model_v1.dat.bz2", a network "trained from scratch on a dataset of about 3 million faces", released with the other models "into the public domain".