How to create tfrecords for VGGFace2 dataset using utils/dataprocess.py.

[Mamtesh11]

Thanks for sharing code for this MobileFaceNet!

I am trying to train this on VGGFace2 dataset, to do that train_nets.py required tfrecords as input data. I am generating tfrecords using below script.

https://github.com/Merle314/WildFace/blob/master/tools/convert_vggface2.py

This script creates multiple shared of tfrecord, I created multiple shared and single shared both and feed to train_nets.py. In the both the cases I am getting below error.

tensorflow.python.framework.errors_impl.InvalidArgumentError: Input to reshape is a tensor with 76800 values, but the requested shape has 37632

I am suspecting above error because of inappropriate format of tfrecord. So, kindly help me to generate correct tfrecords.

Thanks!

[malik-anhar]

Hey @Mamtesh11 I got the same issue. Input to reshape is a tensor with 76800 values, but the requested shape has 37632. Have you found the solution?

[Mamtesh11]

Hi @malik-anhar,

I have overcome from that error after using the tfrecords from VGGFace2@GoogleDrive.

I hope this will help.