[译] Rethinking Image Inpainting via a Mutual Encoder-Decoder with Feature Equalizations

题目：通过具有特征均衡功能的互编解码器重新思考图像修复
作者：Hongyu Liu([email protected]), Bin Jiang, Yibing Song, Wei Huang, and Chao Yang

摘要

简介

程序复现

一、环境搭建

ModuleNotFoundError: No module named 'PIL'

安装pillow

(py37pt13) [intern2@172-10-60-160 Rethinking-Inpainting-MEDFE]$ conda install pillow

ImportError: TensorBoard logging requires TensorBoard with Python summary writer installed. This should be available in 1.14 or above.

用pip豆瓣源安装tensorboard

(py37pt13) [intern2@172-10-60-160 Rethinking-Inpainting-MEDFE]$ pip install tensorboard==1.14 -i http://pypi.douban.com/simple/ --trusted-host=pypi.douban.com/simple

ModuleNotFoundError: No module named 'past'

安装future包

(py37pt11) [intern2@172-10-60-160 Rethinking-Inpainting-MEDFE]$ pip install -i http://pypi.douban.com/simple/ --trusted-host=pypi.douban.com/simple future

FileNotFoundError: [Errno 2] No such file or directory: './logs/Mutual Encoder-Decoder'

补充新建文件夹代码

    if os.path.exists(dir):
        print(dir, " existed and has been removed")
        os.rmdir(dir) 
    os.makedirs(dir, exist_ok=True)

AttributeError: 'Tensor' object has no attribute 'bool'

Pytorch版本升级到1.3.1

二、代码微调
1.查看文件夹下文件个数

(base) [intern2@172-10-60-160 paris_structure]$ ls -l|grep "^-"| wc -l          
14900