pix2pix学习系列（1）：预训练模型测试pix2pix

参考文献：

[Pytorch系列-66]：生成对抗网络GAN - 图像生成开源项目pytorch-CycleGAN-and-pix2pix - 使用预训练模型测试pix2pix模型

运行环境

win 10

1、代码下载

Github

2、下载pix2pix数据集

• 通过URL手工下载：地址

• 存放路径：pytorch-CycleGAN-and-pix2pix\datasets
• 

3、下载预训练模型

• 下载链接
  
• 存放路径
  需要把模型的名称改为latest_net_G.pth，并存放在./checkpoints/facades_label2photo_pretrained目录中。
  

4. 使用Anaconda进行调试

• 打开Anaconda Prompt (Anaconda3)
  

• 激活环境
  activate yolov5_tpz

• 切换到d盘
  输入： d:
  

• 切换到 D:\tpz\the-third-paper\pytorch-CycleGAN-and-pix2pix-master
  输入： cd D:\tpz\the-third-paper\pytorch-CycleGAN-and-pix2pix-master
  

• 运行命令
  输入：
  python test.py --dataroot ./datasets/facades --direction BtoA --model pix2pix --name facades_label2photo_pretrained

• 运行效果

----------------- Options ---------------aspect_ratio: 1.0batch_size: 1checkpoints_dir: ./checkpointscrop_size: 256dataroot: ./datasets/facades                   [default: None]dataset_mode: aligneddirection: BtoA                                 [default: AtoB]display_winsize: 256epoch: latesteval: Falsegpu_ids: 0init_gain: 0.02init_type: normalinput_nc: 3isTrain: False                                [default: None]load_iter: 0                                    [default: 0]load_size: 256max_dataset_size: infmodel: pix2pix                              [default: test]n_layers_D: 3name: facades_label2photo_pretrained       [default: experiment_name]ndf: 64netD: basicnetG: unet_256ngf: 64no_dropout: Falseno_flip: Falsenorm: batchnum_test: 50num_threads: 4output_nc: 3phase: testpreprocess: resize_and_cropresults_dir: ./results/serial_batches: Falsesuffix:use_wandb: Falseverbose: False
----------------- End -------------------
dataset [AlignedDataset] was created
initialize network with normal
model [Pix2PixModel] was created
loading the model from ./checkpoints\facades_label2photo_pretrained\latest_net_G.pth
---------- Networks initialized -------------
[Network G] Total number of parameters : 54.414 M
-----------------------------------------------
creating web directory ./results/facades_label2photo_pretrained\test_latest
D:\Anaconda3\envs\yolov5_tpz\lib\site-packages\torchvision\transforms\transforms.py:280: UserWarning: Argument interpolation should be of type InterpolationMode instead of int. Please, use InterpolationMode enum.warnings.warn(
processing (0000)-th image... ['./datasets/facades\\test\\1.jpg']
processing (0005)-th image... ['./datasets/facades\\test\\103.jpg']
processing (0010)-th image... ['./datasets/facades\\test\\12.jpg']
processing (0015)-th image... ['./datasets/facades\\test\\17.jpg']
processing (0020)-th image... ['./datasets/facades\\test\\21.jpg']
processing (0025)-th image... ['./datasets/facades\\test\\26.jpg']
processing (0030)-th image... ['./datasets/facades\\test\\30.jpg']
processing (0035)-th image... ['./datasets/facades\\test\\35.jpg']
processing (0040)-th image... ['./datasets/facades\\test\\4.jpg']
processing (0045)-th image... ['./datasets/facades\\test\\44.jpg']

• 查看结果
  图片位置： \results\facades_label2photo_pretrained\test_latest\images
  
  

5. 也可以使用pycharm进行调试

• 设置 options/base_options.py

parser.add_argument('--dataroot', default='datasets/facades', help='path to images (should have subfolders trainA, trainB, valA, valB, etc)')
parser.add_argument('--name', type=str, default='facades_label2photo_pretrained', help='name of the experiment. It decides where to store samples and models')
parser.add_argument('--model', type=str, default='pix2pix', help='chooses which model to use. [cycle_gan | pix2pix | test | colorization]')
parser.add_argument('--direction', type=str, default='BtoA', help='AtoB or BtoA')

• 设置 options/test_options.py

parser.set_defaults(model='pix2pix')

注意：
如果不将 parser.set_defaults(model='test ') 更改为 parser.set_defaults(model=‘pix2pix’)，将会出现以下错误：
AttributeError: ‘Sequential’ object has no attribute ‘model’

解决方案参考

• 运行 test.py