normandipalo / faceid_beta

An implementation of iPhone X's FaceID using face embeddings and siamese networks on RGBD images.

faceID_beta

An implementation of iPhone X's FaceID using face embeddings and siamese networks on RGBD images.

You can read more about my implementation details at this link.

The .py file is created automatically from the notebook file, so you could find a strange ordering of code and functions. I highly suggest to use the notebook version.

faceid_beta's Issues

Very nice article

Could you please share the python code as 1 or more files?

So it is a lot easier to try out.

Thanks

Not a zip file

Hello, when I try the code it gives me this error. please help me how to solve it ! Thank you

purpose of [im_in] ?

Hi,
for your cell: [27]
you have: model_top = Model(inputs = [im_in], outputs = feat_x) ..

I'm curious what's the purpose of putting "[ ]" around im_in (since there is only one input) ?
Would : ... Model(inputs = im_in, ... also suffice ?

Thanks

Face ID

contrastive_loss func

Is the contrastive_loss func right?
I think maybe you need to change the postion of (1-y_true) and y_true

cannot reshape array of size 40000 into shape (1,200,200,4)

When I run the code, 433 lines of code get an error::
cannot reshape array of size 40000 into shape (1,200,200,4)
@normandipalo

What's the worry with the .dat data?

hi,
I have got an error when i run this code? see the screenshot

I did some test,And there is some worry at line 481 in .dat file. this code can not break when variable i is 480, and return above error. Is any difference or special about .dat file ending? I will appreciate for any help，thanks

tensorflow version

Hi I'm getting some strange issue using the program. I think that probably the version of tensorflow (cpu on ubuntu 16.04 with python 2.7 on both hardware with I7 and amd phenom) has some trouble. I'm getting some erratic behaviour. Can you please indicate "the best" version of tensorflow to use with the program ?
Thank you for your time.

dataset in the code is missing now

There is no dataset now,
could you help us with this? Thank you.

Can't run python script.

Hi there, thanks for the .py version.

Hope you can help.

When running i get this error:

Traceback (most recent call last):
File ".\faceid.py", line 486, in
model_final.evaluate([cop[0].reshape((1,200,200,1)), cop[1].reshape((1,200,200,1))], np.array([0.]))
File "C:\Users\srei\AppData\Local\Programs\Python\Python36\lib\site-packages\keras\engine\training.py", line 1768, in evaluate
batch_size=batch_size)
File "C:\Users\srei\AppData\Local\Programs\Python\Python36\lib\site-packages\keras\engine\training.py", line 1476, in _standardize_user_data
exception_prefix='input')
File "C:\Users\srei\AppData\Local\Programs\Python\Python36\lib\site-packages\keras\engine\training.py", line 123, in _standardize_input_data
str(data_shape))
ValueError: Error when checking input: expected input_3 to have shape (200, 200, 4) but got array with shape (200, 200, 1)

TypeError: l2_normalize() got an unexpected keyword argument 'axis'

I'm getting the above error raised during batch normalization of x1. The solutions I found are not very encouraging. Can you please suggest something ?
Thanks.