Pix2Pix style conversion. Output the original image to the learned image style. For example: convert black and white to color, edge map to photo, map to satellite map and other applications.
There are train, val, and test in the data/A and data/B folders. A is the image after conversion, and B is the image before conversion.
Please put the images into the train, val, and test in the A and B folders before and after the conversion. One thing to note is that the train, val, and test in the two folders must be able to find the corresponding image file names.
Then click 1. combine A and B to combine the corresponding images into one, and output to data/train, data/val, data/test.
1). It is recommended that the length and width of the image be the same. If they are not the same, you can zoom or crop them.
2). The image file must be .jpg.
3). The images need to be matched in pairs.
Open visdom_server with port 8801.
Open the browser, you can see the loss curve and other information of the training process.
Train the model.
• --continue_train : Load model/latest model to continue training.
• --n_epochs 1000 : The training epochs is 1000.
• --gpu_ids 0 : Which gpu to use for training, you can also set more than one, for example --gpu_ids 0,1,2.
• --save_epoch_freq 5 : Store model frequency.
• --epoch_count 2300: The output model file name starts from 2300 times, which means that there have been 2300 epochs of training before. It can be used with continue_train.
• checkpoints_dir = 'model' : Model folder.
• epoch ='latest': Load the latest_net_G_A model.
• dataroot ='data/B', phase ='test': The test data folder is data/B/test.
• num_test = 27: How many test images are in the folder.
Inferred folder, the images in the folder are the images of A and B combined by running 1_combine_A_and_B.ipynb.
• checkpoints_dir ='model', epoch ='latest': The model is model/latest_net_G_A.pth.
• dataroot ='data/', phase ='test': The test folder is data/test.
• num_test = 20: How many test images are in the folder.
Contact Us and How to Buy
Welcome to contact us. Please refer to the following link: