bruceyang2012 / face-identification-with-cnn-triplet-loss

Face identification with cnn+triplet-loss written by Keras.

Description

This is a implementation of cnn+triplet-loss for face identification written by keras, which is the last step of my FaceID system. You can find another two repositories as follows:

1. Face-detection-with-mobilenet-ssd
2. Face-Alignment-with-simple-cnn
3. Face-identification-with-cnn-triplet-loss

prepare data

1. Download caltech faces from Official Website , and put it into face_data folder in organize_data.py.
2. Run organize_data.py to generate train, test, dev data.
3. Run load_data.py to generate train_x.npy, train_y.npy and so on.

train

Follow face_train.ipynb step by step. You can change the parameters for better performance.

to do

Compare with center loss.

References

meownoid/face-identification-tpe

face-identification-with-cnn-triplet-loss's Issues

Regarding orgnised_data.py and load_data.py

1. After organise_data.py, it blown up 451 images into tons of thousands images, which I think you are trying to increase dataset, that's ok. What I dun understand is why in each subject folders, it contains different faces? Am I correct to say, in real cases, each subject folder should ONLY contain images of that individual? Example: Subject_Bruce folder contains only bruce images... and so on . Same question to Test folder too.

2. Is it mandatory to have each subject folders to have the same number of images?

3. In load_data.py, the generation of train and test npy, estimated to take 20hours. Did u got this too or its just me? I had to modify from float32 to float16, else it will throw memory error.

tpe, tpe_pred = build_tpe(n_in, n_out, W_pca.T)

HI I'm stuck here.
keras merge keep error me out.

In [13]: tpe, tpe_pred = build_tpe(n_in, n_out, W_pca.T)
D:\xrvision\XRV_projects\Face-identification-with-cnn-triplet-loss\model.py:67: UserWarning: Update your Dense call to the Keras 2 API: Dense(18, input_dim=18, weights=[array([[-..., activation="linear", use_bias=False)
base_model.add(Dense(n_out, input_dim=n_in, bias=False, weights=[W_pca], activation='linear'))
Traceback (most recent call last):

File "", line 1, in
tpe, tpe_pred = build_tpe(n_in, n_out, W_pca.T)

File "D:\xrvision\XRV_projects\Face-identification-with-cnn-triplet-loss\model.py", line 75, in build_tpe
e = merge([a_emb, p_emb, n_emb], mode=triplet_merge, output_shape=triplet_merge_shape)

TypeError: 'module' object is not callable

fa = FaceAligner('../model/shape_predictor_68_face_landmarks.dat', '../model/face_template.npy')

@bruceyang2012 ,I want to ask where I can find the file shape_predictor_68_face_landmarks.dat and face_template.npy

No files are being loaded with organize data

Hello, I placed the caltech_faces under face_data. I ran organize_data.py and I get 0 files loaded. What could I be doing wrong?