README.md

metadata

dataset_info:
  features:
    - name: full_colour
      dtype: image
    - name: sketch
      dtype: image
  splits:
    - name: train
      num_bytes: 170204480
      num_examples: 922
  download_size: 170225532
  dataset_size: 170204480

Dataset Card for "one_piece_dataset"

This dataset contains 922 images taken from the one piece anime, with each row containing the coloured image and the sketch one.

Example Setup (the images are not normalized)

transform = Compose([
            transforms.Resize((128, 128)),
            transforms.ToTensor(),
            transforms.Lambda(lambda t: (t * 2) - 1)
])

def train():

def transforms(examples):
    examples["sketch_pixel_values"] = [transform(image.convert("RGB"))  for image in examples["sketch"]]
    examples["full_colour_pixel_values"] = [ transform(image.convert("RGB")) for image in  examples["full_colour"]]
    del examples["sketch"]
    del examples["full_colour"]

    return examples


dataset = load_dataset("pawlo2013/one_piece_dataset", split="train")

transformed_full_colour_dataset = dataset.with_transform(transforms)

dataloader = DataLoader(transformed_full_colour_dataset, batch_size=16, shuffle=True, num_workers=0)