This was developed to demonstrate continuous authentication as a viable authentication mechanism for mobile devices as part of GovernmentWare (GovWare) at Singapore International Cyber Security Week 2017
The app uses Google's Mobile Vision API for Face Detection which is bundled with the front-end. The code is based on this sample code project with minor modifications made for UI.
A pre-trained VGGFace model built with caffe is for Face Recognition and served over REST API as the backend server.
The caffe installation is assumed to be separate, however the code for obtaining the VGG features is part of the project. Also the caffe model used can be found here
- Copy the contents of the vgg model to caffe-root/models/vgg_face_caffe
- Update caffe_root @ projects/views.py to point to the correct path
- Create the folder project/data
- this is where the backend service will store the enrolled data
- Execute
python run.py
-
url: http://localhost:5000/biosecure/api/v1/enroll
request: HTTP PUT multi-part request to be sent along with a json {image: binary_image_file}
response: contains a json reporting the status of enrollment
-
url: http://localhost:5000/biosecure/api/v1/verify
request: HTTP PUT multi-part request to be sent along with a json {image: binary_image_file}
response: contains a json reporting the status. If successfull, it will contain a "score" key with the actual cosine distance Note that cosine distance is a measure of similarity and hence lower values mean better matches!
Sanka Rasnayaka, Divya Sivasankaran
This was not meant to be a standalone project/app. If you are trying to adapt it for your own use case/facing issues with deploying or wish to share your comments/feedback, reach me at div1090@gmail.com.