Closed 252651980 closed 2 years ago
Another question. What is point_model.pt. Is it a training data? Where is the training data if point_model.pt is not in Adeleine.
Thank you for the issue. I will answer your question.
About SketchKeras and superpixel SketchKeras: SketchKeras extracts line arts from colorized images. In order to train the line art colorization network, we need line arts as inputs of the network. Therefore, we have to employ line extraction methods for creating inputs, and SketchKeras is the reasonable one. superpixel: We can divide images into a lot of regions with superpixel algorithms like the figures on the page
About DATA_PATH
, SKETCH_PATH
and SS_PATH
Having done downloading Tagged Anime Illustrations, the path of the dataset corresponds to DATA_PATH
. After this, you need to extract line arts from images in Tagged Anime Illustrations by SketchKeras, and the path containing line arts corresponds to SKETCH_PATH
. As for SS_PATH
, you can set DATA_PATH
or SKETCH_PATH
as SS_PATH
because superpixel images in SS_PATH
are not actually used in atari_userhint_v2
.
About point_model.pt
point_model.pt
is not the training data. It is the file containing weights of the trained model.
Thanks a lot for answering my questions. Where should I place the training data on Adeleine if it's ok from atari_userhint_v2.
I do not get the point. Adeleine
is a GUI application that enables users to test their line arts. Training data is not needed on Adeleine, test data and the trained model file are needed.
I do not get the point.
Adeleine
is a GUI application that enables users to test their line arts. Training data is not needed on Adeleine, test data and the trained model file are needed.
I am sorry. My English is not that good. What I mean is how to associate Adeleine and atari_userhint_v2
I should use atari_userhint_v2 to train the data, which pictures downloaded, to get a file like print_model.pt. Then put it into the dir './Adeleine/ckpts/'. Is it right?
If you train your own model by using your own dataset, it is correct
Okay. Thanks a lot.
Hi, thanks for your resource, It's really fantastic!
I am a noob about python. I have some question about atari_userhint_v2:
Thanks.