From 3ffd0b82ff8770e41da200f1e1bf383f163ecb2e Mon Sep 17 00:00:00 2001
From: yangbin <136039852@qq.com>
Date: Thu, 14 Jan 2021 03:30:42 +0800
Subject: [PATCH 1/4] =?UTF-8?q?=E4=BF=AE=E6=AD=A3=E9=83=A8=E5=88=86?=
=?UTF-8?q?=E6=96=87=E4=BB=B6=EF=BC=8C=E4=B8=BA=E5=B7=A5=E7=A8=8B=E5=A2=9E?=
=?UTF-8?q?=E5=8A=A0=E8=AF=B4=E6=98=8E?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
---
.idea/vcs.xml | 6 +
.idea/workspace.xml | 146 +++++++++++++-----
README.md | 26 +++-
data_utils.py | 1 +
model.py | 1 +
psnrmeter.py | 1 +
.../SRF_3/{comic.png => compare_comic.png} | Bin
test_image.py | 1 +
train.py | 1 +
9 files changed, 143 insertions(+), 40 deletions(-)
create mode 100644 .idea/vcs.xml
rename results/SRF_3/{comic.png => compare_comic.png} (100%)
diff --git a/.idea/vcs.xml b/.idea/vcs.xml
new file mode 100644
index 0000000..94a25f7
--- /dev/null
+++ b/.idea/vcs.xml
@@ -0,0 +1,6 @@
+
+
+
+
+
+
\ No newline at end of file
diff --git a/.idea/workspace.xml b/.idea/workspace.xml
index 9413d3d..e6ddd55 100644
--- a/.idea/workspace.xml
+++ b/.idea/workspace.xml
@@ -1,7 +1,13 @@
-
+
+
+
+
+
+
+
@@ -15,12 +21,12 @@
-
+
-
-
+
+
@@ -28,23 +34,35 @@
-
-
+
+
-
-
+
+
-
+
-
-
+
+
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
@@ -52,12 +70,28 @@
+
+
+
+
+
+
+
+
+
+
+
+
@@ -70,9 +104,9 @@
-
-
-
+
+
+
@@ -81,7 +115,6 @@
-
@@ -89,20 +122,16 @@
-
-
-
-
-
+
-
+
@@ -197,30 +226,31 @@
-
+
+
-
-
+
+
-
+
-
+
-
+
-
-
-
+
+
+
@@ -233,7 +263,7 @@
-
+
@@ -243,23 +273,67 @@
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
-
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/README.md b/README.md
index cf35d62..7dec1fc 100644
--- a/README.md
+++ b/README.md
@@ -1,6 +1,24 @@
# ESPCN_Learning
***
-### 1、本地Anaconda4.9.2,pytorch1.4,python3.7.7
-### 2、项目工程用Pycharm创建
-### 3、本项目所用数据集为VOC2012
-
+## 介绍
+ 超分辨重建的经典论文《Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network》,首次提出了用亚像素卷积层对图片进行上采样操作,相比与SRCNN的整体对图片进行上采样操作后再进行超分辨重建,通过使用亚像素卷积极大的缩小了运算量,重建时间以及提高了重建质量。
+***
+## 项目环境
+### 1、本地Anaconda4.9.2,pytorch1.4,python3.7.7
+### 2、项目工程用Pycharm创建
+### 3、本项目所用数据集为VOC2012,由于数据集较大,未放入仓库中。数据集获取地址: https://pan.baidu.com/s/1xDoBr28KIFaEEU_uIcgfSQ 提取密码:oyql
+***
+## 操作说明
+### 1、下载好VOC2012数据集后添加至工程的data文件夹中;
+### 2、在Pycharm中运行 data_utils.py ,这一步是为了创建训练所需的数据集;
+### 3、运行 train.py 即可开始训练,训练的epoch可根据需要调整,文件夹epochs用来保存最棒的训练结果;
+### 4、test_image.py 用于测试训练好的网络。
+***
+## 结果展示
+** 在3倍放大因子的条件下进行的重建,所用训练结果为/epochs/ESPCN_best(5).pt **
+### butterfly.png
+![avatar](/results/SRF_3/compare_butterfly.png)
+** PSNR: 25.016115 (db), SSIM: 0.855480 **
+### comic.png
+![avatar](/results/SRF_3/compare_comic.png)
+** PSNR: 23.070899 (db), SSIM: 0.758211 **
\ No newline at end of file
diff --git a/data_utils.py b/data_utils.py
index 4745fff..cf507ad 100644
--- a/data_utils.py
+++ b/data_utils.py
@@ -1,3 +1,4 @@
+# -*- coding: UTF-8 -*-
import argparse
import os
from os import listdir
diff --git a/model.py b/model.py
index bdfa313..bc3a018 100644
--- a/model.py
+++ b/model.py
@@ -1,3 +1,4 @@
+# -*- coding: UTF-8 -*-
import matplotlib.pyplot as plt
import torch
import torch.nn as nn
diff --git a/psnrmeter.py b/psnrmeter.py
index edc083c..8c8d826 100644
--- a/psnrmeter.py
+++ b/psnrmeter.py
@@ -1,3 +1,4 @@
+# -*- coding: UTF-8 -*-
import math
from math import log10
import numpy as np
diff --git a/results/SRF_3/comic.png b/results/SRF_3/compare_comic.png
similarity index 100%
rename from results/SRF_3/comic.png
rename to results/SRF_3/compare_comic.png
diff --git a/test_image.py b/test_image.py
index 6c1f785..0a7ddee 100644
--- a/test_image.py
+++ b/test_image.py
@@ -1,3 +1,4 @@
+# -*- coding: UTF-8 -*-
import argparse
import os
from os import listdir
diff --git a/train.py b/train.py
index 923b9d5..0314dd3 100644
--- a/train.py
+++ b/train.py
@@ -1,3 +1,4 @@
+# -*- coding: UTF-8 -*-
import argparse
import copy
--
Gitee
From d7114ac4772711d9f8aaa25f6502f82397cec86d Mon Sep 17 00:00:00 2001
From: vegee <136039852@qq.com>
Date: Thu, 14 Jan 2021 03:33:28 +0800
Subject: [PATCH 2/4] =?UTF-8?q?update=20README.md.=20=E4=BF=AE=E6=94=B9?=
=?UTF-8?q?=E6=A0=BC=E5=BC=8F?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
---
README.md | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/README.md b/README.md
index 7dec1fc..dbc9692 100644
--- a/README.md
+++ b/README.md
@@ -1,7 +1,7 @@
# ESPCN_Learning
***
## 介绍
- 超分辨重建的经典论文《Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network》,首次提出了用亚像素卷积层对图片进行上采样操作,相比与SRCNN的整体对图片进行上采样操作后再进行超分辨重建,通过使用亚像素卷积极大的缩小了运算量,重建时间以及提高了重建质量。
+ *超分辨重建的经典论文《Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network》,首次提出了用亚像素卷积层对图片进行上采样操作,相比与SRCNN的整体对图片进行上采样操作后再进行超分辨重建,通过使用亚像素卷积极大的缩小了运算量,重建时间以及提高了重建质量。*
***
## 项目环境
### 1、本地Anaconda4.9.2,pytorch1.4,python3.7.7
@@ -15,10 +15,10 @@
### 4、test_image.py 用于测试训练好的网络。
***
## 结果展示
-** 在3倍放大因子的条件下进行的重建,所用训练结果为/epochs/ESPCN_best(5).pt **
+**在3倍放大因子的条件下进行的重建,所用训练结果为/epochs/ESPCN_best(5).pt**
### butterfly.png
![avatar](/results/SRF_3/compare_butterfly.png)
-** PSNR: 25.016115 (db), SSIM: 0.855480 **
+**PSNR: 25.016115 (db), SSIM: 0.855480**
### comic.png
![avatar](/results/SRF_3/compare_comic.png)
-** PSNR: 23.070899 (db), SSIM: 0.758211 **
\ No newline at end of file
+**PSNR: 23.070899 (db), SSIM: 0.758211**
\ No newline at end of file
--
Gitee
From 8f6984c7cba2330dabe3ad18d553f6fd7bda6578 Mon Sep 17 00:00:00 2001
From: vegee <136039852@qq.com>
Date: Thu, 14 Jan 2021 03:37:04 +0800
Subject: [PATCH 3/4] =?UTF-8?q?update=20README.md.=20=E7=AC=AC=E4=BA=8C?=
=?UTF-8?q?=E6=AC=A1=E8=B0=83=E6=95=B4=E6=A0=BC=E5=BC=8F?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
---
README.md | 13 +++++++------
1 file changed, 7 insertions(+), 6 deletions(-)
diff --git a/README.md b/README.md
index dbc9692..a21adb7 100644
--- a/README.md
+++ b/README.md
@@ -1,12 +1,13 @@
# ESPCN_Learning
***
## 介绍
- *超分辨重建的经典论文《Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network》,首次提出了用亚像素卷积层对图片进行上采样操作,相比与SRCNN的整体对图片进行上采样操作后再进行超分辨重建,通过使用亚像素卷积极大的缩小了运算量,重建时间以及提高了重建质量。*
+ **超分辨重建的经典论文*《Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network》*,首次提出了用亚像素卷积层对图片进行上采样操作,相比与SRCNN的整体对图片进行上采样操作后再进行超分辨重建,通过使用亚像素卷积极大的缩小了运算量,重建时间以及提高了重建质量。**
***
## 项目环境
### 1、本地Anaconda4.9.2,pytorch1.4,python3.7.7
### 2、项目工程用Pycharm创建
-### 3、本项目所用数据集为VOC2012,由于数据集较大,未放入仓库中。数据集获取地址: https://pan.baidu.com/s/1xDoBr28KIFaEEU_uIcgfSQ 提取密码:oyql
+### 3、本项目所用数据集为VOC2012,由于数据集较大,未放入仓库中。
+ 数据集获取地址: https://pan.baidu.com/s/1xDoBr28KIFaEEU_uIcgfSQ 提取密码:oyql
***
## 操作说明
### 1、下载好VOC2012数据集后添加至工程的data文件夹中;
@@ -17,8 +18,8 @@
## 结果展示
**在3倍放大因子的条件下进行的重建,所用训练结果为/epochs/ESPCN_best(5).pt**
### butterfly.png
-![avatar](/results/SRF_3/compare_butterfly.png)
-**PSNR: 25.016115 (db), SSIM: 0.855480**
+![avatar](/results/SRF_3/compare_butterfly.png)
+ **PSNR: 25.016115 (db), SSIM: 0.855480**
### comic.png
-![avatar](/results/SRF_3/compare_comic.png)
-**PSNR: 23.070899 (db), SSIM: 0.758211**
\ No newline at end of file
+![avatar](/results/SRF_3/compare_comic.png)
+ **PSNR: 23.070899 (db), SSIM: 0.758211**
\ No newline at end of file
--
Gitee
From a8d25f6aea9cd3fe134fdd9800a2eb388a550987 Mon Sep 17 00:00:00 2001
From: vegee <136039852@qq.com>
Date: Thu, 14 Jan 2021 03:39:56 +0800
Subject: [PATCH 4/4] =?UTF-8?q?=E7=AC=AC=E4=B8=89=E6=AC=A1=E4=BF=AE?=
=?UTF-8?q?=E6=94=B9?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
---
README.md | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/README.md b/README.md
index a21adb7..067567c 100644
--- a/README.md
+++ b/README.md
@@ -1,7 +1,7 @@
# ESPCN_Learning
***
## 介绍
- **超分辨重建的经典论文*《Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network》*,首次提出了用亚像素卷积层对图片进行上采样操作,相比与SRCNN的整体对图片进行上采样操作后再进行超分辨重建,通过使用亚像素卷积极大的缩小了运算量,重建时间以及提高了重建质量。**
+ **超分辨重建的经典论文** *《Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network》* **,首次提出了用亚像素卷积层对图片进行上采样操作,相比与SRCNN的整体对图片进行上采样操作后再进行超分辨重建,通过使用亚像素卷积极大的缩小了运算量,重建时间以及提高了重建质量。**
***
## 项目环境
### 1、本地Anaconda4.9.2,pytorch1.4,python3.7.7
@@ -19,7 +19,7 @@
**在3倍放大因子的条件下进行的重建,所用训练结果为/epochs/ESPCN_best(5).pt**
### butterfly.png
![avatar](/results/SRF_3/compare_butterfly.png)
- **PSNR: 25.016115 (db), SSIM: 0.855480**
+**PSNR: 25.016115 (db), SSIM: 0.855480**
### comic.png
![avatar](/results/SRF_3/compare_comic.png)
- **PSNR: 23.070899 (db), SSIM: 0.758211**
\ No newline at end of file
+**PSNR: 23.070899 (db), SSIM: 0.758211**
\ No newline at end of file
--
Gitee