sketchyscene / sketchyscenecolorization Goto Github PK

It seems from an initial look through of the various readme files that the pipeline also requires the segment_data_generation.py script from the SketchyScene repository to perform an initial preprocessing step on the sketches. Currently this means that both repositories need to be utilized... Are there any plans to integrate this into the SketchySceneColorization repository in the future?

Also, it appears that the Foreground Instance Colorization module is designed to handle individual instances for inference rather than being able to take the output of the Instance Matching module. This seems to be missing a step where the instances matched are separated and individually put through the Foreground Instance Colorization module. Are there any plans to add this functionality?

Finally, the Background Colorization module seems to be lacking an inference method. While it should be possible to utilize the test method for the same purpose, are there plans to implement an inference method to handle example data?

Traceback (most recent call last):
File "/content/SketchySceneColorization/Instance_Matching/matching_main.py", line 523, in
captions_base_dir=args.captions_base_dir)
File "/content/SketchySceneColorization/Instance_Matching/matching_main.py", line 141, in train
print('firstly train, loaded', ckpt.model_checkpoint_path)
AttributeError: 'NoneType' object has no attribute 'model_checkpoint_path'

Hello I faced this problem and I dont know how to solve it

Traceback (most recent call last):
File "/content/SketchySceneColorization/Instance_Matching/matching_main.py", line 523, in
captions_base_dir=args.captions_base_dir)
File "/content/SketchySceneColorization/Instance_Matching/matching_main.py", line 141, in train
print('firstly train, loaded', ckpt.model_checkpoint_path)
AttributeError: 'NoneType' object has no attribute 'model_checkpoint_path'

I have downloaded the resnet_pretrained_model and put the files in the diractory named SketchyScene_DeepLabv2