search

rcmalli / keras-vggface

VGGFace implementation with Keras Framework
MIT License
928 stars 417 forks source link

Which version of ResNet-50 train was converted? #51

Open vitoralbiero opened 5 years ago

vitoralbiero commented 5 years ago

On the VGGFace2 paper, the authors have two versions of ResNet-50, one trained from scratch on VGGFace2 dataset, and other trained on MS-Celeb-1M and fine-tuned on VGGFace2. Which one was converted to Keras and is in this repository?

Thank you!

zhangdaoxian commented 5 years ago

On the VGGFace2 paper, the authors have two versions of ResNet-50, one trained from scratch on VGGFace2 dataset, and other trained on MS-Celeb-1M and fine-tuned on VGGFace2. Which one was converted to Keras and is in this repository?

Thank you!

I noticed that the output size of the top layer is 8631, which is the number of classes of the training set of VGGFace2, so I guess this one the trained solo on VGGFace2

riobintang commented 2 weeks ago

On the VGGFace2 paper, the authors have two versions of ResNet-50, one trained from scratch on VGGFace2 dataset, and other trained on MS-Celeb-1M and fine-tuned on VGGFace2. Which one was converted to Keras and is in this repository? Thank you!

I noticed that the output size of the top layer is 8631, which is the number of classes of the training set of VGGFace2, so I guess this one the trained solo on VGGFace2

Couldn't we train it on MS Celeb and then modify the last layer to fine tune on VGGFace2? need explanation about ResNet-50