thank you! Next thing is to work on resolution - Jeremy talks low res -> hi res in part 2.
tbh i just skimmed part2 so far, but i tried CycleGAN for data augmentation and found it so fascinating that decided to apply it to my hobbies which are cooking and sketching
the CycleGAN project page is quite comprehensive, and the PyTorch code worked like a charm…
more fun with cycleGAN - it’s so engrossing - keep starring at the screen mesmerized by the training process
the installment above was all about fusing my mixology photos with my sketches (stills and flowers)
this one is the intermediate results of training CycleGAN on my food pix and food sketches…
when i started this project, the idea was to get some insights on my cooking/drawing style but in the process i got so taken by the generated imagery - colors, textures and line work, that now i want to go way deeper into the creative side of DL…
Can you share your working?
Thanks in Advance…
thank you! the code is pretty much PyTorch CycleGAN cloned from the original github repo
the input and output are all hosted on my flickr:
GenerativeArt
Note to self: Wait for my GPU system and not spend AWS Credits on extremely cool applications yet
i’m lucky to have a GTX 1080 at home - still it takes 2 days to train cycleGAN - can’t imagine doing this on AWS
My college friends area worried about me being always seen watching youtube tutorials or reading books. They are concerned when do I study DL.
Little do they know I have a AWS server running quietly, printing errors at the rate of 1 per 5 minutes
2 days!
I was discussing RL with @lgvaz, an Atari game from the 90s takes 4 days to train on his GTX-1070!