Facenet: ValueError while training triplet loss

[PoonamZ]

@davidsandberg I am facing the ValueError when I am training the triplet loss. Please can you suggest the solution?

python src/align/align_dataset_mtcnn.py /home/poonam/Downloads/Bright_IMG /home/poonam/Downloads/Bright_IMG1 --image_size 182 --margin 44
Creating networks and loading parameters
Traceback (most recent call last):
File "src/align/align_dataset_mtcnn.py", line 161, in
main(parse_arguments(sys.argv[1:]))
File "src/align/align_dataset_mtcnn.py", line 57, in main
pnet, rnet, onet = align.detect_face.create_mtcnn(sess, None)
File "/home/poonam/projects/facenet/src/align/detect_face.py", line 283, in create_mtcnn
pnet.load(os.path.join(model_path, 'det1.npy'), sess)
File "/home/poonam/projects/facenet/src/align/detect_face.py", line 85, in load
data_dict = np.load(data_path, encoding='latin1').item() #pylint: disable=no-member
File "/home/poonam/.virtualenvs/facenet/lib/python3.6/site-packages/numpy/lib/npyio.py", line 447, in load
pickle_kwargs=pickle_kwargs)
File "/home/poonam/.virtualenvs/facenet/lib/python3.6/site-packages/numpy/lib/format.py", line 692, in read_array
raise ValueError("Object arrays cannot be loaded when "
ValueError: Object arrays cannot be loaded when allow_pickle=False

[thelastfunction]

check if your model_path is correct and det1.npy exists in directory
It seems your mtcnn face detection model is not loaded correctly in line pnet.load(os.path.join(model_path, 'det1.npy'), sess)

[PoonamZ]

@thelastfunction Thanks.

[deepkakkad3]

@PoonamZ How You Solve that??

[doyanger]

I have the same issue on line 283 of the file detect_face.py,
I also add a print before to confirm is the model_path is correct, it is OK.
the det1.npy, det2.npy, det2.npy just right in the folder of /src/align/

There are also many warnings, indicating some useage of code is deprecated!

`W0622 18:15:39.963512 139750941083392 deprecation_wrapper.py:119] From src/align/align_dataset_mtcnn.py:52: The name tf.GPUOptions is deprecated. Please use tf.compat.v1.GPUOptions instead.

W0622 18:15:39.964641 139750941083392 deprecation_wrapper.py:119] From src/align/align_dataset_mtcnn.py:53: The name tf.Session is deprecated. Please use tf.compat.v1.Session instead.

W0622 18:15:39.964763 139750941083392 deprecation_wrapper.py:119] From src/align/align_dataset_mtcnn.py:53: The name tf.ConfigProto is deprecated. Please use tf.compat.v1.ConfigProto instead.

2019-06-22 18:15:39.965052: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
WARNING: Logging before flag parsing goes to stderr.
W0622 18:15:39.969522 139792992794368 deprecation_wrapper.py:119] From src/align/align_dataset_mtcnn.py:52: The name tf.GPUOptions is deprecated. Please use tf.compat.v1.GPUOptions instead.

W0622 18:15:39.971852 139792992794368 deprecation_wrapper.py:119] From src/align/align_dataset_mtcnn.py:53: The name tf.Session is deprecated. Please use tf.compat.v1.Session instead.

W0622 18:15:39.973070 139792992794368 deprecation_wrapper.py:119] From src/align/align_dataset_mtcnn.py:53: The name tf.ConfigProto is deprecated. Please use tf.compat.v1.ConfigProto instead.

2019-06-22 18:15:39.973934: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
WARNING: Logging before flag parsing goes to stderr.
W0622 18:15:39.978578 139680673736448 deprecation_wrapper.py:119] From src/align/align_dataset_mtcnn.py:52: The name tf.GPUOptions is deprecated. Please use tf.compat.v1.GPUOptions instead.

WARNING: Logging before flag parsing goes to stderr.
W0622 18:15:39.979246 139680673736448 deprecation_wrapper.py:119] From src/align/align_dataset_mtcnn.py:53: The name tf.Session is deprecated. Please use tf.compat.v1.Session instead.

W0622 18:15:39.979369 139680673736448 deprecation_wrapper.py:119] From src/align/align_dataset_mtcnn.py:53: The name tf.ConfigProto is deprecated. Please use tf.compat.v1.ConfigProto instead.

2019-06-22 18:15:39.979740: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2019-06-22 18:15:39.980931: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2112000000 Hz
W0622 18:15:39.977033 140020089046784 deprecation_wrapper.py:119] From src/align/align_dataset_mtcnn.py:52: The name tf.GPUOptions is deprecated. Please use tf.compat.v1.GPUOptions instead.

W0622 18:15:39.982066 140020089046784 deprecation_wrapper.py:119] From src/align/align_dataset_mtcnn.py:53: The name tf.Session is deprecated. Please use tf.compat.v1.Session instead.

W0622 18:15:39.982222 140020089046784 deprecation_wrapper.py:119] From src/align/align_dataset_mtcnn.py:53: The name tf.ConfigProto is deprecated. Please use tf.compat.v1.ConfigProto instead.
.....
`

[thelastfunction]

@doyanger you may be using tensorflow 2.0 version, thats why you are getting these warnings ......you can ignore them ... or you can downgrade it to tensorflow 1.8

[mohammadkhaleghy]

Hi Guys can someone help me with this I get the same error
detect_face.py
print("this is data path="+os.path.join(model_path, 'det1.npy')) pnet.load(os.path.join(model_path, 'det1.npy'), sess)

the det1.npy file is existed in the directory, still I get this error

this is data path=/Users/mohammadkhaleghi/AI/facenet/src/align/det1.npy Traceback (most recent call last): File "align_dataset_mtcnn.py", line 160, in <module> main(parse_arguments(sys.argv[1:])) File "align_dataset_mtcnn.py", line 56, in main pnet, rnet, onet = detect_face.create_mtcnn(sess, None) File "/Users/mohammadkhaleghi/AI/facenet/src/align/detect_face.py", line 284, in create_mtcnn pnet.load(os.path.join(model_path, 'det1.npy'), sess) File "/Users/mohammadkhaleghi/AI/facenet/src/align/detect_face.py", line 85, in load data_dict = np.load(data_path, encoding='latin1').item() #pylint: disable=no-member File "/opt/anaconda3/lib/python3.7/site-packages/numpy/lib/npyio.py", line 453, in load pickle_kwargs=pickle_kwargs) File "/opt/anaconda3/lib/python3.7/site-packages/numpy/lib/format.py", line 722, in read_array raise ValueError("Object arrays cannot be loaded when " ValueError: Object arrays cannot be loaded when allow_pickle=False

[soniritesh124]

File "alignalign_dataset_mtcnn.py", line 170, in
main(args)
File "alignalign_dataset_mtcnn.py", line 53, in main
pnet, rnet, onet = align.detect_face.create_mtcnn(sess, None)
File "C:UsersRitesh's LaptopDesktopfacial-recognition-video-facenetsrcaligndetect_face.py", line 283, in create_mtcnn
pnet.load(os.path.join(model_path, 'det1.npy'), sess)
File "C:UsersRitesh's LaptopDesktopfacial-recognition-video-facenetsrcaligndetect_face.py", line 85, in load
data_dict = np.load(data_path, encoding='latin1').item() #pylint: disable=no-member
File "C:UsersRitesh's LaptopAppDataLocalProgramsPythonPython36libsite-packagesnumpylibnpyio.py", line 447, in load
pickle_kwargs=pickle_kwargs)
File "C:UsersRitesh's LaptopAppDataLocalProgramsPythonPython36libsite-packagesnumpylibformat.py", line 692, in read_array
raise ValueError("Object arrays cannot be loaded when "
ValueError: Object arrays cannot be loaded when allow_pickle=False

[KowsalyaR97]

@soniritesh124 This may due do the version complexity in numpy, what numpy version are you using?

[us]

They didn't give a specific version of package so these are broke.
Please try install this versions. Problem will solved.

pip install numpy==1.16
pip install scipy==1.1.0

[lkrazy]

@us
Thank you, it works now