It was my groups' final assignment of course deep learning, which mixed several exited GANs to create a GAN of our own. It was used on a shoes dataset to generate shoes images
- down load dataset
- change config in config.py
- run
python my_code.py
- python >= 3.6
- pytorch >= 1.0
- baiduyun link: https://pan.baidu.com/s/1oxmfGmgAwPoxkY1R5Yk6_g
password: xeww
- 12 classes dataset in [data1], 5 classes dataset in [data2]
- network parameters
- my_dataset.py
- imsize and num_classes in config.py
配置文件在 config.py里,数据集下载后修改config.py里面的数据集路径和batchsize之类的配置 然后python my_code.py 就可以了
python 3.6以上 pytorch 1.0以上,若想在GPU上运行需要配置好pytorch的GPU版本
链接: https://pan.baidu.com/s/1oxmfGmgAwPoxkY1R5Yk6_g
提取码: xeww
若想在自己的数据集上运行,只需要修改my_dataset.py 即可, 对应imsize和分类数在config.py中修改 evaluate 可以参照my_tool.py