GuichenSociety / stylegan2-tiny

It is more concise than stylegan2. If you want to learn stlegan network, you can come in and learn it.比stylegan2更简洁,如果是想学习stlegan网络可以进来学习一下。
4 stars 0 forks source link

stylegan2-tiny

It is more concise than stylegan2. If you want to learn stlegan network, you can come in and learn it.比stylegan2更简洁,如果是想学习stlegan网络可以进来学习一下。

image

Installation

Install project

conda create -n stylegan2-tiny python 3.x
conda activate stylegan2-tiny
cd stylegan2-tiny
pip install -r requirements.txt

Train

Enter the config file and modify it.进入config文件下进行修改。

Epochs = 1000                                    Training times.训练次数。
Image_size = 64                                  Enter or generate picture size.输入或生成图片大小。
Dataset_path = "dataset_human"                   Data set address, subfolder containing pictures.数据集地址,子文件夹包含图片。
Image_format = "jpg"                             
Batch_size = 16                                  Training batch of data set.数据集的训练批次。
Load_weights = True

Execute the file train.py.执行文件train.py。

python train.py

Test

Under the test() function, num_ IMG is to select and generate several pictures.在test()函数下,num_img是选择生成几张图片。

python test.py

🔗 Links

Referenced articles.参考的文章。 stylegan3