数据增强和转换
固定转换
- 边缘补充像素(Pad)
- 尺寸变换(Resize)
- 中心截取(CenterCrop)
- 顶角及中心截取(FiveCrop)
- 尺灰度变换(GrayScale)
概率控制的转换
- 随机垂直翻转(RandomVerticalFlip)
- 随机应用(RandomApply)
# -*- coding: utf-8 -*-
from PIL import Image
from pathlib import Path
import matplotlib.pyplot as plt
import numpy as np
import torch
import torchvision.transforms as T
plt.rcParams["savefig.bbox"] = 'tight'
orig_img = Image.open('images.jpg')
torch.manual_seed(0)
def plot(imgs, title, with_orig=True, row_title=None, **imshow_kwargs):
if not isinstance(imgs[0], list):
imgs = [imgs]
num_rows = len(imgs)
num_cols = len(imgs[0]) + with_orig
fig, axs = plt.subplots(nrows=num_rows, ncols=num_cols, squeeze=False)
plt.title(title)
for row_idx, row in enumerate(imgs):
row = [orig_img] + row if with_orig else row
for col_idx, img in enumerate(row):
ax = axs[row_idx, col_idx]
ax.imshow(np.asarray(img), **imshow_kwargs)
ax.set(xticklabels=[], yticklabels=[], xticks=[], yticks=[])
if with_orig:
axs[0, 0].set(title='Original image')
axs[0, 0].title.set_size(8)
if row_title is not None:
for row_idx in range(num_rows):
axs[row_idx, 0].set(ylabel=row_title[row_idx])
plt.tight_layout()
# 边缘补充
padded_imgs = [T.Pad(padding=padding)(orig_img) for padding in (3, 10, 30, 50)]
plot(padded_imgs, "T.pad")
# 尺寸变换
resized_imgs = [T.Resize(size=size)(orig_img) for size in (30, 50, 100, orig_img.size)]
plot(resized_imgs, title='Resize')
# 中心截取
center_crops = [T.CenterCrop(size=size)(orig_img) for size in (30, 50, 100, orig_img.size)]
plot(center_crops, title='CenterCrop')
# 四角及中间截取
(top_left, top_right, bottom_left, bottom_right, center) = T.FiveCrop(size=(100, 100))(orig_img)
plot([top_left, top_right, bottom_left, bottom_right, center], title='FiveCrop')
# 灰度变换
gray_img = T.Grayscale()(orig_img)
plot([gray_img], cmap='gray', title='Grayscale')
# 颜色抖动转换
jitter = T.ColorJitter(brightness=.5, hue=.3)
jitted_imgs = [jitter(orig_img) for _ in range(4)]
plot(jitted_imgs, title='ColorJitter')
# 高斯模糊
blurrer = T.GaussianBlur(kernel_size=(5, 9), sigma=(0.1, 5))
blurred_imgs = [blurrer(orig_img) for _ in range(4)]
plot(blurred_imgs, title='GaussianBlur')
# 随机透视变换
perspective_transformer = T.RandomPerspective(distortion_scale=0.6, p=1.0)
perspective_imgs = [perspective_transformer(orig_img) for _ in range(4)]
plot(perspective_imgs, title='RandomPerspective')
# 随机旋转
rotater = T.RandomRotation(degrees=(0, 180))
rotated_imgs = [rotater(orig_img) for _ in range(4)]
plot(rotated_imgs, title='RandomRotation')
# 随机仿射变换
affine_transfomer = T.RandomAffine(degrees=(30, 70), translate=(0.1, 0.3), scale=(0.5, 0.75))
affine_imgs = [affine_transfomer(orig_img) for _ in range(4)]
plot(affine_imgs, title='RandomAffine')
# 弹性变换
elastic_transformer = T.ElasticTransform(alpha=250.0)
transformed_imgs = [elastic_transformer(orig_img) for _ in range(2)]
plot(transformed_imgs, title='ElasticTransform')
# 随机裁剪
cropper = T.RandomCrop(size=(128, 128))
crops = [cropper(orig_img) for _ in range(4)]
plot(crops, title='RandomCrop')
# 随机缩放裁剪
resize_cropper = T.RandomResizedCrop(size=(32, 32))
resized_crops = [resize_cropper(orig_img) for _ in range(4)]
plot(resized_crops, title='RandomResizedCrop')
# 随机颜色翻转
inverter = T.RandomInvert()
invertered_imgs = [inverter(orig_img) for _ in range(4)]
plot(invertered_imgs, title='RandomInvert')
# 随机海报化
posterizer = T.RandomPosterize(bits=2)
posterized_imgs = [posterizer(orig_img) for _ in range(4)]
plot(posterized_imgs, title='RandomPosterize')
# 随机调节锐利度
sharpness_adjuster = T.RandomAdjustSharpness(sharpness_factor=2)
sharpened_imgs = [sharpness_adjuster(orig_img) for _ in range(4)]
plot(sharpened_imgs, title='RandomAdjustSharpness')
# 随机调节对比度
autocontraster = T.RandomAutocontrast()
autocontrasted_imgs = [autocontraster(orig_img) for _ in range(4)]
plot(autocontrasted_imgs, title='RandomAutocontrast')
# 随机直方图均衡
equalizer = T.RandomEqualize()
equalized_imgs = [equalizer(orig_img) for _ in range(4)]
plot(equalized_imgs, title='RandomEqualize')
augmenter = T.RandAugment()
imgs = [augmenter(orig_img) for _ in range(4)]
plot(imgs, title='RandAugment')
# 随机垂直翻转
hflipper = T.RandomHorizontalFlip(p=0.5)
transformed_imgs = [hflipper(orig_img) for _ in range(4)]
plot(transformed_imgs, title='RandomHorizontalFlip')
# 随机水平翻转
vflipper = T.RandomVerticalFlip(p=0.5)
transformed_imgs = [vflipper(orig_img) for _ in range(4)]
plot(transformed_imgs, title='RandomVerticalFlip')
# 随机应用
applier = T.RandomApply(transforms=[T.RandomCrop(size=(64, 64))], p=0.5)
transformed_imgs = [applier(orig_img) for _ in range(4)]
plot(transformed_imgs, title='RandomApply')
plt.show()
综合案例
# -*- coding: utf-8 -*-
import os
import cv2 as cv
import torch
from torch.utils.data import Dataset, DataLoader
import numpy as np
from torchvision import transforms
from torch.nn import Sequential
transform = Sequential( # 生成一个一系列的操作
transforms.GaussianBlur(kernel_size=(5, 9), sigma=(0.1, 5)),
transforms.RandomVerticalFlip(p=0.5),
transforms.RandomHorizontalFlip(p=0.2)
)
class MyDataset(Dataset):
def __init__(self):
root_data = "dataset"
self.file_name_list = []
for root, dirs, files in os.walk(root_data):
for file_i in files:
file_i_full_path = os.path.join(root, file_i)
self.file_name_list.append(file_i_full_path)
def __len__(self):
return len(self.file_name_list)
def __getitem__(self, item):
file_i_loc = self.file_name_list[item]
image_i = cv.imread(file_i_loc)
image_i = cv.resize(image_i, dsize=(256, 256))
image_i = np.transpose(image_i, (2, 0, 1))
image_i_tensor = torch.from_numpy(image_i)
image_i_tensor = transform(image_i_tensor)
file_i_loc_info = file_i_loc.split('\\')
file_i_loc_info[0] = new_root
new_file_i_loc = os.path.join(file_i_loc_info[0], file_i_loc_info[1], file_i_loc_info[2])
return image_i_tensor, new_file_i_loc
if __name__ == '__main__':
new_root = 'my_new_dataset'
my_dataset = MyDataset()
dataloader = DataLoader(my_dataset)
for x_i, loc_i in dataloader:
x_i=x_i.view(3,256,256)
loc_info = loc_i[0].split('\\')
file_dir = os.path.join(loc_info[0], loc_info[1])
print(file_dir)
if os.path.isdir(file_dir):
pass
else:
os.makedirs(file_dir)
image = transforms.ToPILImage()(x_i)
image.save(loc_i[0])
