有用的文章參考：
https://github.com/jwyang/faster-rcnn.pytorch
https://github.com/tiancity-NJU/da-faster-rcnn-PyTorch
https://github.com/divyam02/dafrcnn-pytorch
https://blog.csdn.net/xinyu_cheng/article/details/89321655
https://github.com/naoto0804/cross-domain-detection
https://github.com/VisionLearningGroup/DA_Detection
https://www.cnblogs.com/wzyuan/p/11105758.html
https://www.commonlounge.com/discussion/da0c2181dbd34c4aa7776f0a351408b8
https://zhuanlan.zhihu.com/p/57737230
#########################################################################
在服務器上新建立用戶，然後安裝anaconda
bash Anaconda3-5.2.0-Linux-x86_64.sh
conda create -n tch04 python=3.6
conda activate tch04
conda install pytorch=0.4.0 cuda90 -c pytorch
conda install torchvision=0.2.1
conda install cudatoolkit=9.0
conda update -n base -c defaults conda
conda update numpy
pip install numpy==1.17.0
pip install --upgrade numpy
conda install scipy=1.2.1
========================================================================
查看驅動
nvidia-smi
安裝cuda9.0
sudo sh cuda_9.0.176_384.81_linux.run --no-opengl-libs
Do you accept the previously read EULA?
2 accept/decline/quit: accept
3 Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 384.81?
4 (y)es/(n)o/(q)uit: n
5 Install the CUDA 9.0 Toolkit?
6 (y)es/(n)o/(q)uit: y
7 Enter Toolkit Location
8 [ default is /usr/local/cuda-9.0 ]:
9 Do you want to install a symbolic link at /usr/local/cuda?
10 (y)es/(n)o/(q)uit: y
11 Install the CUDA 9.0 Samples?
12 (y)es/(n)o/(q)uit: y
13 Enter CUDA Samples Location
14 [ default is /home/pertor ]:
15 Installing the CUDA Toolkit in /usr/local/cuda-9.0 ...
16 Missing recommended library: libXmu.so
sudo vim ~/.bashrc
“i”
export PATH=/usr/local/cuda-9.0/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda-9.0/lib64:$LD_LIBRARY_PATH
“esc” “：wq：”
source ~/.bashrc
檢查是否安裝成功
cd /usr/local/cuda-9.0/samples/1_Utilities/deviceQuery
sudo make
./deviceQuery
安裝cudnn
cd -
tar -zxvf cudnn-9.0-linux-x64-v7.6.2.24.tgz
shenfanshu@ubuntu-server:~/A_ment$ cd /usr/local/cuda
shenfanshu@ubuntu-server:/usr/local/cuda$ ls
bin include libnsight nvml samples tools
doc jre libnvvp nvvm share version.txt
extras lib64 nsightee_plugins pkgconfig src
shenfanshu@ubuntu-server:/usr/local/cuda$ sudo rm -rf /usr/local/cuda
shenfanshu@ubuntu-server:/usr/local/cuda$ cd ..
shenfanshu@ubuntu-server:/usr/local$ sudo rm -rf /usr/local/cuda
shenfanshu@ubuntu-server:/usr/local$ ls
bin cuda-9.0 games lib python share
cuda-8.0 etc include man sbin src
shenfanshu@ubuntu-server:/usr/local$ sudo ln -s /usr/local/cuda-9.0 /usr/local/cuda
shenfanshu@ubuntu-server:/usr/local$ ls
bin cuda-8.0 etc include man sbin src
cuda cuda-9.0 games lib python share
shenfanshu@ubuntu-server:/usr/local/cuda$ cd /home/shenfanshu/A_ment
shenfanshu@ubuntu-server:~/A_ment$
shenfanshu@ubuntu-server:~/A_ment$ sudo cp cuda/include/cudnn.h /usr/local/cuda-9.0/include/
shenfanshu@ubuntu-server:~/A_ment$ sudo cp cuda/lib64/libcudnn* /usr/local/cuda-9.0/lib64/ -d
shenfanshu@ubuntu-server:~/A_ment$ sudo chmod a+r /usr/local/cuda-9.0/include/cudnn.h
shenfanshu@ubuntu-server:~/A_ment$ sudo chmod a+r /usr/local/cuda-9.0/lib64/libcudnn*
shenfanshu@ubuntu-server:~/A_ment$ nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2017 NVIDIA Corporation
Built on Fri_Sep__1_21:08:03_CDT_2017
Cuda compilation tools, release 9.0, V9.0.176
========================================================================
測試torch0.4.0的安裝
(tch04) shenfanshu@ubuntu-server:~/B_code/faster-rcnn.pytorch-master$ python
Python 3.6.9 |Anaconda, Inc.| (default, Jul 30 2019, 19:07:31)
[GCC 7.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
>>> torch.__version__
'0.4.0'
>>> torch.version.cuda
'9.0.176'
>>> quit()
========================================================================
測試例程的運行
#############################
faster-rcnn.pytorch-master
-----------------------------
簡單的faster-rcnn，已經事先參照官網步驟下載好了所有需要使用到的數據集
torch = 0.4.0
python = 3.6
torchvision=0.2.1
cuda = 9.0
#############################
(tch04) shen@ubuntu-server:~/myments/anaconda/doc$ cd /home/shen/C_codeTemplate/faster-rcnn.pytorch-master
(tch04) shen@ubuntu-server:~/C_codeTemplate/faster-rcnn.pytorch-master$ pip install -r requirements.txt
(tch04) shen@ubuntu-server:~/C_codeTemplate/faster-rcnn.pytorch-master$ cd lib
(tch04) shen@ubuntu-server:~/C_codeTemplate/faster-rcnn.pytorch-master/lib$ sh make.sh
(tch04) shen@ubuntu-server:~/C_codeTemplate/faster-rcnn.pytorch-master/lib$ cd ..
(tch04) shen@ubuntu-server:~/C_codeTemplate/faster-rcnn.pytorch-master$
###訓練###
CUDA_VISIBLE_DEVICES=2 python trainval_net.py --dataset pascal_voc --net vgg16 --bs 1 --nw 4 --cuda
[session 1][epoch 1][iter 2100/10022] loss: 0.9306, lr: 1.00e-03
fg/bg=(14/242), time cost: 17.289636
rpn_cls: 0.1146, rpn_box: 0.0270, rcnn_cls: 0.2293, rcnn_box 0.1312
###批量測試###
python test_net.py --dataset pascal_voc --net vgg16 --checksession 1 --checkepoch 20 --checkpoint 10021 --cuda
###demo###
(tch04) shen@ubuntu-server:~/C_codeTemplate/faster-rcnn.pytorch-master
python demo.py --net vgg16 --checksession 1 --checkepoch 20 --checkpoint 10021 --cuda
會生成det結尾的圖片
========================================================================
例程代碼運行2
da-faster-rcnn-PyTorch-master
-----------------------------
簡單的faster-rcnn，已經事先參照官網步驟下載好了所有需要使用到的數據集
#############################
cd /home/shen/C_codeTemplate/da-faster-rcnn-PyTorch-master
(tch04) shenfanshu@ubuntu-server:~/B_code/da-faster-rcnn-PyTorch-master$ pip install -r requirements.txt
(tch04) shenfanshu@ubuntu-server:~/B_code/da-faster-rcnn-PyTorch-master$ cd lib(tch04) shenfanshu@ubuntu-server:~/B_code/da-faster-rcnn-PyTorch-master/lib$ sh make.sh
(tch04) shenfanshu@ubuntu-server:~/B_code/da-faster-rcnn-PyTorch-master/lib$ cd ..
#-----------------------------------------
# train cityscapes -> cityscapes-foggy
#-----------------------------------------
CUDA_VISIBLE_DEVICES=0 python da_trainval_net.py --dataset cityscape --net vgg16 --bs 1 --lr 2e-3 --lr_decay_step 6 --cuda
allow_unreachable=True) # allow_unreachable flag
RuntimeError: cuda runtime error (2) : out of memory at /opt/conda/conda-bld/pytorch_1524586445097/work/aten/src/THC/generic/THCStorage.cu:58
超內存了，確實運行的程序有點多了不行，哈哈哈
驚訝的發現服務器有4塊1080,一直用的是第0塊，可以用第1塊，哈哈哈
(base) shenfanshu@ubuntu-server:~/B_code/da-faster-rcnn-PyTorch-master$ conda activate tch04
(tch04) shenfanshu@ubuntu-server:~/B_code/da-faster-rcnn-PyTorch-master$ CUDA_VISIBLE_DEVICES=1 python da_trainval_net.py --dataset cityscape --net vgg16 --bs 1 --lr 2e-3 --lr_decay_step 6 --cuda
############################################################################
遠程連接服務器，關閉終端離線運行程序
############################################################################
首先，創建你自己的screen：
screen -S 你的screen名字
我在這裏輸入的是 screen -S ShenfanshuCode。
然後，當你想看看程序跑的咋樣了，可以用如下命令查看服務器當前運行的screen：
screen -ls
可以看到這臺服務器上運行着兩個程序，12115.ShenfanshuCode這個就是我剛剛創建的screen
(base) shenfanshu@ubuntu-server:~$ screen -ls
There is a screen on:
12115.ShenfanshuCode (2019年09月22日 09時11分07秒) (Attached)
1 Socket in /var/run/screen/S-shenfanshu.
再輸入如下命令重新鏈接：
screen -r 12115.ShenfanshuCode
回來了~
當你不需要這個screen時，直接輸入 kill 12115 就ok了。
############################################################################
# 給pycharm 自動註釋
############################################################################
"""
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# @project : ${PROJECT_NAME}
# @Time : ${DATE} ${TIME}
# @Author : twinkle
# @File : ${NAME}.py
# @Software: ${PRODUCT_NAME}
"""
############################################################################
安裝jupyter notebook
https://www.cnblogs.com/honway/p/9559324.html
https://blog.csdn.net/weixin_40321858/article/details/89487375
############################################################################
1.打開命令行，重新生成一個jupyter配置文件：
jupyter notebook --generate-config
修個配置文件，找到這個新生成的文件 jupyter_notebook_config.py
(base) twinkle@twinkle-ubuntu:~$ conda activate tch04
(tch04) twinkle@twinkle-ubuntu:~$ cd /home/twinkle/Myments/anaconda3/envs/tch04
(tch04) twinkle@twinkle-ubuntu:~/Myments/anaconda3/envs/tch04$ jupyter notebook --generate-config
Writing default config to: /home/twinkle/.jupyter/jupyter_notebook_config.py
(tch04) twinkle@twinkle-ubuntu:~/Myments/anaconda3/envs/tch04$ gedit /home/twinkle/.jupyter/jupyter_notebook_config.py
(tch04) twinkle@twinkle-ubuntu:~/Myments/anaconda3/envs/tch04$ vim /home/twinkle/.jupyter/jupyter_notebook_config.py
(tch04) twinkle@twinkle-ubuntu:~/Myments/anaconda3/envs/tch04$ sudo gedit /home/twinkle/.jupyter/jupyter_notebook_config.py
[sudo] twinkle 的密碼：
2.搜索 NotebookApp.allow_password_change，改爲：NotebookApp.allow_password_change=False ，記得去掉註釋的#3.回到windows命令行，運行jupyter notebook password
(base) twinkle@twinkle-ubuntu:~$ jupyter notebook password
Traceback (most recent call last):
File "/home/twinkle/Myments/anaconda3/bin/jupyter", line 7, in <module>
from jupyter_core.command import main
ModuleNotFoundError: No module named 'jupyter_core'
(base) twinkle@twinkle-ubuntu:~$ conda activate tch04
(tch04) twinkle@twinkle-ubuntu:~$ jupyter notebook password
Enter password: 1234..[]
Verify password:
[NotebookPasswordApp] Wrote hashed password to /home/twinkle/.jupyter/jupyter_notebook_config.json
(tch04) twinkle@twinkle-ubuntu:~$ sudo gedit /home/twinkle/.jupyter/jupyter_notebook_config.json
[sudo] twinkle 的密碼：
{
"NotebookApp": {
"password": "sha1:085ece0d1dd1:553443e277c35932f210c940014394220d25157a"
}
}
#密碼生成的一串sha1，寫入到了這個文件，等下要把這串sha1，複製到第一個生成的配置文件中
4.還差一步密碼纔可以生效使用，在第一個jupyter_notebook_config.py配置文件中找到“c.NotebookApp.password“，等於，剛生成的那個密碼sha1，效果如下：去掉前面的”#“
c.NotebookApp.password =u'sha1:085ece0d1dd1:553443e277c35932f210c940014394220d25157a'
5.保險起見，把jupyter重啓，之後，無論是更換瀏覽器還是Logout登陸，新密碼都可以正常使用了。
6. 將anaconda3的相關環境與包關聯到jupyter notebook上。然後網頁 conda 選項卡下就會顯示anaconda中的已經安裝的環境，選中環境，還可以在下方看到該環境已經安裝的包，並且可以直接給環境安裝需要的包
conda install nb_conda
7. 顯示anaconda中已經安裝的虛擬環境，過程中保證打開jupyter-notebook的終端不要關閉
activate 環境名稱
conda install ipykernel
python -m ipykernel install --user --name 環境名稱 --display-name "Python (環境名稱)"
2019年9月20日