Could you please provide the download link for the pretrained model "top_model.h5"? Thanks!

Hi,
Please can you provide pre-trained model ?

hey! I've been following your code, great work!

I assume you run all cells in order.

The only mistake I see is that, first, you apply the data augmentation technique, then, you split your data in train and test. This means that you may be adding an example on train set and its augmented version to the test set. This may lead to overestimated performance on your test set. You need to make sure that Train and Test sets are as independent as possible.

The easy fix is, fist split in train test, and then apply augmentation on both datasets.

The provided Dropbox url is not working anymore, please provide one that actually works.

Hello, I am not a computer science person, but I have interest in having an offline usable image to font detection app on my computer. If there was a instruction guide to show users how to setup this program, it would be very helpful.

For example, I have "Font_Rec(DeepFont).ipynb" opened inside Jupyter Notebook but I don't know what to do to get things going.
Where do I upload the unknown font image? And do I need to give the the program these libraries to get it working? My goal is to have an offline usable font detection app.

Hi, thank you so much for the code. I see there are 4 types of augmentation in the code ("blur","noise","affine","gradient"). Are these two types of augmentations (Variable Character Spacing, Variable Aspect Ratio) implemented? Thanks!

As the paper said that Cu layers are train unsupervised in type CAE first, then they will be imported to the full models and freeze. Later, the Cs layers are trained. But in the notebook, I do not see that progress, may I misunderstand the paper or something that I may miss.
Thank you so much for your code

What are the versions of the libraries used? Doesn't work in current version. Thanks.

Hi, I've tried to implement the DeepFont paper multiple times but I'm unable to reproduce the results. I was curious if you or anyone was able to get them font classifier to work on real / natural data, as opposed to the synthetic generated data. I was able to get good accuracy on the synthetic data fine.