Open lilianabrandao opened 4 years ago
Hey
I am wondering about the exact same thing. When fine-tuning the model or, in my case, stacking dense layers on the 'pool5' layer, should I use the preprocessing_input function or should I do preprocessing based on my own dataset?
I want to use the vggface weights for training an emotion classifier. Subtracting the mean of the vgg16 dataset from my face images seems weird.
Thanks
Hi @steinela
"Subtracting the mean of the vgg16 dataset from my face images seems weird." I felt exactly the same way but it seems to be the right thing to do. At least that's the intuition I get from some VGG Transfer Learning implementations I saw on the internet (e.g. https://discuss.pytorch.org/t/vgg-transfer-learning/33808).
Please share any additional information you feel may be helpful.
Thank you.
Hi @rcmalli
First of all, thank you for sharing this amazing work/repository.
I'm using transfer learning to fine-tune VGG-Face (2015 model).
I know I have to apply the same image pre-processing to my training images as in the original paper (i.e. "The input to all networks is a face image of size 224×224 with the average face image (computed from the training set) subtracted"), but I run into a doubt: should I use the average face image of the original training dataset or should I use the average face image of my training dataset?
I've tried to find the answer but without success. Any clues? Thanks