Shu Kong

Let me upload some code for training this weekend:-) On Thu, Apr 25, 2019 at 7:13 AM zlw21gxy wrote: > Hey, thanks for your great paper, I wonder if there...

Hey, I've uploaded the script to train the models. By referring the following link, you should be able to run. https://github.com/aimerykong/predictive-filter-flow/blob/master/mgPFF_video/demo00_training.ipynb You can also prepare your own data by poking...

Hi, thanks for your interest! We have not thought of this yet. As PFF generates per-pixel kernels warping input for the output, instead of producing the output directly, the input...

Hi, Thanks for your interest. I didn't provide the list of image pairs because the txt file that stores the pairs is >200MB, meaning millions pairs are sampled for testing...

Hi, As we don't have the right to host the AVA data, we can't publish a public link. However, we do have a copy of AVA dataset. It is large,...

You should merge the parts before unzip ref. https://superuser.com/questions/15935/how-do-i-reassemble-a-zip-file-that-has-been-emailed-in-multiple-parts On Tue, Jul 6, 2021 at 7:46 AM Stergiadis Manos ***@***.***> wrote: > I downloaded those 3 parts but cannot unzip...

hi, Yuan, Thank you for your interest! When using caffe to train models, people usually have prototxt files for training and testing/application separately. Here I didn't include the training prototxt...

When training the attribute branches, I fixed the model and only learned new layers for the new branches. In practice, like you guessed (2), I trained each attribute branch at...

Great to see it works on your end! It's true that one might get better or worse results due to some randomness, like data preparation. I notice that some people...

Hello, Jiyuan, Thanks for trying all these. When I did this project, I first trained for binary classification on the aesthetic label by fine-tuning Alexnet. Then over the classification model,...