DeecampW1_Carla&Tensorflow

1. Carla Simulator

    编译包安装

tar -zxvf ./CARLA_0.8.4.tar.gz
cd CarlaUE4/Binaries/Linux
./CarlaUE4 #启动虚拟环境

    地图切换

./CarlaUE4 /Game/Maps/Town01
or
./CarlaUE4 /Game/Maps/Town02

   教程原址

 

2. 批量文件重命名 自动编号

     i=1; for x in *; do mv $x $i.png; let i=i+1; done

     i 的值为起始编号

 

3. 将文件夹所有内容复制到另一个文件夹

    eg. 将MyCamera文件夹下所有文件复制到mask

    cp -rf ./MyCamera/* ../mask/

 

4. 从7000个数中随机抽取5600个作为训练集，再从剩下的1400个数中抽取700个作为检验集，剩下的作为测试集。

    随机抽取样本采用 random.sample (数据集，抽取量)

    求取两个数组的差集，set (数组A) ^ set (数组B)

import random

values = range(1,7001)
trainSet = random.sample(values,5600)
rest = set(values)^set(trainSet)
evalSet = random.sample(rest,700)
visSet = set(evalSet)^set(rest)

fp = open('train.txt','w+')
for line in trainSet:
 fp.write(str(line) + '\n')
fp.close()

fp = open('eval.txt','w+')
for line in evalSet:
 fp.write(str(line) + '\n')
fp.close()

fp = open('vis.txt','w+')
for line in visSet:
 fp.write(str(line) + '\n')
fp.close()

5. Tensorflow 安装  

官网教程

• Use pip in a virtual environment (recommended)

将依赖包放在当前目录

export PYTHONPATH=$PYTHONPATH:`pwd`:`pwd`/slim

将依赖包添加到环境变量 vim ~/.bashrc

export PYTHONPATH=$PYTHONPATH:/home/chenyuting/DeeplabV3+/models/research:/home/chenyuting/DeeplabV3+/models/research/slim

安装好之后在虚拟环境下启动tensorflow

source ~/tensorflow/venv/bin/activate      # bash, sh, ksh, or zsh

6. DeepLab V3+ 跑 Cityscapes数据

https://blog.csdn.net/u011974639/article/details/80574306

 

7. DeepLab V3+ 跑自己数据

https://blog.csdn.net/u011974639/article/details/80948990