InsightFace
DeepInsightOpen-source face-analysis project covering face detection, alignment, recognition and attributes, known for the ArcFace recognition loss and the RetinaFace and SCRFD detectors. The insightface Python package runs its pretrained model packs through ONNX Runtime and downloads the default buffalo_l pack on first use; the repository also holds training code for detection and recognition. The library code is MIT-licensed, and the pretrained models are for non-commercial research, with commercial model licenses sold separately.
Openness
2 medium confidence- license
- MIT(library and training code, stated in the README
- source
- public(github.com/deepinsight/insightface)
- core features withheld
- yes — commercial licenses for the recognition model packs and the InspireFace SDK are sold on request
The library code is MIT-licensed, but the face models it downloads by default are licensed for non-commercial research only, and a business that wants to deploy them has to buy a commercial model license from InsightFace. The repository carries no root LICENSE file; the terms are stated in the README and in signed license files beside each model pack. InsightFace-Model-License covers a part the product requires and does not permit commercial use, so the license tier is noncommercial.
- https://raw.githubusercontent.com/deepinsight/insightface/HEAD/python-package/docs/commercial_evaluation.md recorded 2026-09-27
"Commercial Evaluation Mode" document, which walks through evaluating on the user's own data "before choosing a commercial model license" and closes with "Commercial Licensing Next Steps".
- https://raw.githubusercontent.com/deepinsight/insightface/HEAD/python-package/README.md recorded 2026-09-27
Python package README: "The library code is released under the **MIT License**, for academic and commercial use. **The pretrained models provided with this library are for non-commercial research only**, whether downloaded automatically or manually."; "FaceAnalysis() defaults to buffalo_l and downloads the model package".
- https://raw.githubusercontent.com/deepinsight/insightface/HEAD/README.md recorded 2026-09-27
README: "The code of InsightFace is released under the MIT License. There is no limitation for both academic and commercial usage."; "The training data containing the annotation (and the models trained with these data) are available for non-commercial research purposes only."; "For open-sourced face recognition models (e.g., buffalo_l package), please contact recognition-oss-pack@insightface.ai for licensing."
- https://raw.githubusercontent.com/deepinsight/insightface/HEAD/server/backend/insightface_server/licensing/defaults/buffalo_l/MODEL.LICENSE recorded 2026-09-27
buffalo_l MODEL.LICENSE: "license_id": "buffalo_l-public-v1", "issuer": "InsightFace", "grant": "non-commercial".
- https://ungh.cc/repos/deepinsight/insightface/files/master recorded 2026-09-27
Repository tree of 2,029 paths with no LICENSE, LICENSE.txt or LICENSE.md at the root; per-model MODEL.LICENSE files under server/backend/insightface_server/licensing/defaults/ and recognition/arcface_torch training code.
Adoption
3 high confidenceMeasured on monthly PyPI downloads of the insightface package.
- https://pypistats.org/api/packages/insightface/recent recorded 2026-09-27
last_month 932,087 downloads of insightface
Capability
4 medium confidenceInsightFace detects and aligns faces in any photo and recognizes people the user enrolls by comparing embeddings, so a new set of identities needs no training run. It covers one object class, faces, where Ultralytics spans many tasks.
- https://raw.githubusercontent.com/deepinsight/insightface/HEAD/python-package/README.md recorded 2026-09-27
Python package README: "InsightFace provides face detection, recognition, alignment, and attributes through Python and ONNX Runtime."; example "app = FaceAnalysis()" with the buffalo_l pack downloaded on first use.
Verified 2026-09-27