A tensorflow implementation from xsr-ai about mobilefacenet with pretrained parameters
The usage of this repo is same as Xsr-ai's implementation. Please review the page.
The pretrained model is transformed from the model provided by insightface in mxnet version. The network architectures provided by Xsr-ai and insightface are different, so I mainly modify the architecture of Xsr-ai’s implementation.
The pretrained model can be downloaded by here.
You can test the pretrained model by run "python verification.py" when you have all dependencies and data.
This is my log when I verificate my code:
[lfw]XNorm: 11.375188
[lfw]Accuracy: 0.00000+-0.00000
[lfw]Accuracy-Flip: 0.99450+-0.00478
Max of [lfw] is 0.99450