https://github.com/doubangotelecom/huaweifaceliveness

Open Source Face Liveness detection using Huawei ML Kit (Anti-Spoofing)
https://github.com/doubangotelecom/huaweifaceliveness

anti-spoofing face-detection face-liveness face-recognition huawei liveness-detection mlkit spoofing

Last synced: about 1 month ago
JSON representation

Open Source Face Liveness detection using Huawei ML Kit (Anti-Spoofing)

• Host: GitHub
• URL: https://github.com/doubangotelecom/huaweifaceliveness
• Owner: DoubangoTelecom
• License: bsd-3-clause
• Created: 2021-04-18T23:05:49.000Z (about 4 years ago)
• Default Branch: main
• Last Pushed: 2021-04-19T05:05:56.000Z (about 4 years ago)
• Last Synced: 2025-03-21T05:21:48.267Z (about 1 month ago)
• Topics: anti-spoofing, face-detection, face-liveness, face-recognition, huawei, liveness-detection, mlkit, spoofing
• Language: Java
• Homepage:
• Size: 1.29 MB
• Stars: 14
• Watchers: 3
• Forks: 9
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
This repo contains source code using [Huawei's ML Kit for Liveness detection](https://developer.huawei.com/consumer/en/doc/development/HMSCore-Guides/liveness-detection-0000001051386243).

Our tests show **32.2% accuracy for Huawei on high resolution images while [Doubango's Liveness detection](https://www.doubango.org/webapps/face-liveness/) has 100% accuracy**:

  - Tests done on #50 4K images and #50 2K images

  - The images are displayed on MacBook Pro (Retina, 15-inch, Mid 2014)

  - The images are captured using a Galaxy S10+

  - The Galaxy S10+ is positioned at 3 different positions for each image: far, near, close

  - A sample image can be found [here](selfie.jpg). Huawei never managed to catch Print-Attacks on 4K images

Huawei's implementation may not be accurate enough for commercial applications but the [SDK is free](https://developer.huawei.com/consumer/en/doc/development/HMSCore-Guides/ml-service-billing-0000001051010023) and could be good candidate for open source projects.

The next [video](https://youtu.be/52W2K5yJIl4) ([https://youtu.be/4Z8VRTS8WrA](https://youtu.be/52W2K5yJIl4)) shows some tries using ML Kit (code in this repo). 



[![Print-Attack on ML Kit (anti-spoofing)](thumbnail.png)](https://youtu.be/52W2K5yJIl4)