pix2pix输入图片转化为pix2pixhd的格式的脚本

最新推荐文章于 2024-05-06 17:50:45 发布

Xx collider

最新推荐文章于 2024-05-06 17:50:45 发布

阅读量606

点赞数

分类专栏： python 文章标签： python

本文链接：https://blog.csdn.net/mermerno/article/details/122859623

版权

python 专栏收录该内容

2 篇文章 0 订阅

订阅专栏

默认参数下pix2pixhd在较小同分辨率（256,512）下效果也比pix2pix好,该脚本适用pix2pix tf官方和pix2pixhd pytorch官方code的数据集格式转化

from PIL import Image
import os 
path = "/Users/xxxxx/"
os.listdir(path)
to_path_A = '/Users/xxxx/A'
to_path_B = '/Users/xxxx/B'

#分割图片
def cut_image(image,count):
  dir = path + image
  dir = Image.open(dir)
  width, height = dir.size[0],dir.size[1]
  item_width = int(width / count) 
  item_height = height
  box_list = []
    # (left, upper, right, lower)#(0,0,256,256)(256,0,512,256)
  for i in range(0,count):
    box = (i*item_width,0,(i+1)*item_width,item_height)
    box_list.append(box)
  #左A右B保存图片
  image_list = [dir.crop(box) for box in box_list]
  for i in range(len(image_list)):
    if i == 0:
      trainB = image_list[0]
      trainB.save(to_path_B + image, 'PNG')
    if i == 1:
      trainA = image_list[1]
      trainA.save(to_path_A + image, 'PNG')
  return image +'---done---'

for image in os.listdir(path):
  if not image.endswith('png'):
    continue
  cut_image(image,2)