You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Thanks for sharing this valuable repo.
I have concerns about the system requirement. What is the required RAM? and on which operating system does it work the best?
I would like to train the network for my dataset. I only take 40 images for TrainingSetPreparation.py and reduced the batch size to 2, on a Linux OS with 16 GB RAM, the program gets killed during the first or the second epoch.
Training on the mentioned datasets (LFPW, HELEN, IBUGS, etc.) is also not possible.
I also tested on a windows system and the result is the same.
I installed py2.7 and the libraries according to ReadMe.
I look forward to seeing your answer.
The text was updated successfully, but these errors were encountered:
Hello,
sorry for the late reply. There are not specific requirements, but keep in mind that this method was designed and implemented more than 5 years ago so any modern system should be fine (preferrably one equipped with a GPU).
One thing you may find challenging is installing versions of theano and lasagne that are compatible with modern GPUs as those libraries are not supported for quite a while now.
Also, training on 40 images is unlikely to generalize well, you will likely need a lot more than that, at least hundreds.
Thanks for sharing this valuable repo.
I have concerns about the system requirement. What is the required RAM? and on which operating system does it work the best?
I would like to train the network for my dataset. I only take 40 images for TrainingSetPreparation.py and reduced the batch size to 2, on a Linux OS with 16 GB RAM, the program gets killed during the first or the second epoch.
Training on the mentioned datasets (LFPW, HELEN, IBUGS, etc.) is also not possible.
I also tested on a windows system and the result is the same.
I installed py2.7 and the libraries according to ReadMe.
I look forward to seeing your answer.
The text was updated successfully, but these errors were encountered: