PACS数据集

Paper：Self-supervised Domain Adaptation for Computer Vision Tasks

GitHub：https://github.com/robertofranceschi/Domain-adaptation-on-PACS-dataset

数据集下载：https://github.com/MachineLearning2020/Homework3-PACS/tree/master/PACS

PACS数据集总共9991张图片，每张图片3x227x227
7 classes：Dog, Elephant, Giraffe, Guitar, Horse, House, Person
4 domains: Art painting, Cartoon, Photo and Sketch.
Photo (1,670 images), Art Painting (2,048 images), Cartoon (2,344 images) and Sketch (3,929 images)

在这里插入图片描述

用Pytorch加载PACS数据集

详见：https://github.com/ValerioDiEugenio/DomainAdaptation-PACSDataset/blob/main/DomainAdaptation.ipynb

class Pacs(VisionDataset):
    def __init__(self, root, split='', transform=None, target_transform=None):
        
        
        self.split = split
        self.root = os.path.join(root,self.split)
        #print(self.root)
        self.transform = transform
        

        self.dataset= []

        self.dataset = ImageFolder(self.root, transform=transform, loader=pil_loader)
        print(self.dataset[2])
        print(len(self.dataset))

    def __getitem__(self, index):
      
      return self.dataset[index]

    def __len__(self):
      return len(self.dataset)

最后以dog类别为例，用Python代码展示四种不同风格的图片

Python可视化图片数据集的代码

修改dir_path为对应的文件夹，dir_path = f"{DATA_PATH}/PACS/art_painting/dog"

import os
import matplotlib.pyplot as plt
import random
from PIL import Image


def plotPics(data, h=3, w=3, filename="out.jpg"):
    fig, ax_array = plt.subplots(h, w, figsize=(15, 15))

    axes = ax_array.flatten()

    for i, ax in enumerate(axes):
        ri = random.randint(0, len(data) - 1)
        ax.imshow(data[ri], cmap=plt.cm.gray)

    plt.setp(axes, xticks=[], yticks=[], frame_on=False)
    fig.tight_layout()
    fig.savefig(filename)
    plt.show()


DATA_PATH = os.path.abspath(os.path.join(os.path.dirname(__file__), "../data/raw_data"))

dir_path = f"{DATA_PATH}/PACS/art_painting/dog"

data = []
for pic in os.listdir(dir_path):
    data.append(Image.open(f"{dir_path}/{pic}"))

plotPics(data, h=5, w=5)