Skip to content

Open source 2D/3D face and body analysis library with toolkit for ML model retraining and generalization improvement.

License

Notifications You must be signed in to change notification settings

Cvartel/Open-Source-Face-SDK

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Cvartel Open Source SDK

Cvartel SDK is an open-source library that analyzes images to identify human faces, bodies and skeletal joints. The library is top rated in NIST FRVT with 99.62% face recognition accuracy and is available for C++, Python, C#, and Java programming languages on Windows and Linux. It contains Open Model Improvement and Domain Adaptation Scripts which allows you to improve model performance for your specific tasks.

Face Detection and Landmarks Face Recognition Face Attributes
face_detector face_detector face_detector
Detect Faces and Face Landmarks Create and Match Face Templates Estimate age, gender, emotions and much more
Body Detection and Human Pose Estimation
face_detector
Detect Human Bodies and Estimate Skeleton Keypoints

See the list of available models.

Open MIDAS

Beside SDK this repository provides Open Model Improvement and Domain Adaptation Scripts. This toolkit contains the set of techniques and pipelines for models retraining and generalization improvement.
Check out BatchNorms training technique - our first published method (very easy to use fine-tune technique, +2%TPR@1e-6 for pretrained models form insightface repo).

Contributions

Improvements to the SDK and Open MIDAS are always welcome. Please fork the repository and get started. Feel free to open an issue if you want to

  • report a bug
  • submit a fix
  • discuss current state of this repo
  • propose a new feature

License

Affero GPLv3 Commercial-friendly licensing available. Contact us at info@cvartel.com.