radiumray/ubuntuGettingStart

ubuntu上手说明

临时笔记

安装openssh:

sudo apt-get install openssh-server

sudo /etc/init.d/ssh stop

sudo /etc/init.d/ssh start

配置免密码：

https://www.linuxidc.com/Linux/2016-12/139018.htm

sudo visudo

开机自动启动：

https://blog.csdn.net/time_future/article/details/85805298

sudo nano /etc/profile

最下边添加：

python3 ~/carTest/showIP.py &

机器人各种算法：

https://atsushisakai.github.io/PythonRobotics/

yolo数据集 https://www.kaggle.com/aruchomu/data-for-yolo-v3-kernel

c++ opencv win: https://blog.csdn.net/zhaiax672/article/details/88971248

https://colab.research.google.com/drive/1mirG8BSoB3k87mh-qyY3-8-ZXj0XB6h6#scrollTo=Xh5cMw23YdaE

---

import numpy as np import cv2

img = np.zeros([50,50,3],dtype=np.uint8)

sss=np.ones([50, 50, 1], dtype=np.uint8) sss=sss*200

img[:, :, 0] = sss[:, :, 0] img[:, :, 1] = sss[:, :, 0] img[:, :, 2] = sss[:, :, 0]

img=cv2.normalize(img, None, 0, 255, cv2.NORM_MINMAX, cv2.CV_8U)

cv2.imshow('sss', img) cv2.waitKey(0)

https://zhuanlan.zhihu.com/p/29567314

---

给电脑装显卡驱动：

关掉securtiy boot mode 先设置密码，然后disable

去英伟达官网下驱动： https://www.geforce.cn/drivers

ubunut查gpu显卡型号：

lspci | grep -i nvidia

GeForce MX150 64bit

sudo bash NVIDIA-Linux-x86_64-430.50.run

nvida-smi

sudo apt-get purge nvidia*
sudo apt-get --purge remove xserver-xorg-video-nouveau
sudo apt-cache search nvidia | grep -E "nvidia-[0-9]{3}"
sudo apt-get update
sudo apt-get install gcc make


# 查看是否有nouveau驱动
lsmod | grep nouveau
# 如果有输出表示没有仅用，需要禁用

# 如何有效的禁用 Ubuntu18.04的nouveau驱动

sudo nano /etc/modprobe.d/blacklist.conf
# 在最后两行添加：
blacklist nouveau

options nouveau modeset=0
# 禁用nouveau的第三方驱动，之后也不需要改回来

# 执行刷新内核
sudo update-initramfs -u

sudo reboot

#重启后确认：lsmod | grep nouveau 如果没有输出表示成功




chmod +x NVIDIA-Linux-x86_64-418.56.run
sudo ./NVIDIA-Linux-x86_64-418.56.run
sudo  shutdown -r now
nvidia-smi


--------------------------
更新 or  安装驱动

# 查看已有驱动
cat /proc/driver/nvidia/version

# 检查最新驱动并推荐
ubuntu-drivers devices

# 自动安装驱动
sudo ubuntu-drivers autoinstall

# 查看是否有nouveau驱动
lsmod | grep nouveau
# 如果有输出表示没有仅用，需要禁用

# 如何有效的禁用 Ubuntu18.04的nouveau驱动
sudo nano /etc/modprobe.d/blacklist.conf

# 在最后两行添加：
blacklist nouveau

# 禁用nouveau的第三方驱动，之后也不需要改回来
options nouveau modeset=0


# 重启
sudo reboot

nvidia-smi

NVIDIA-SMI has failed because it couldn’t communicate with the NVIDIA driver. Ma ke sure that the latest NVIDIA driver is installed and running

遇到这个问题解决办法： 方法一： cd /usr/src 查看驱动版本号(我的是410.93)

ls 查看你的驱动版本型号

sudo apt-get install dkms sudo dkms install -m nvidia -v 410.93 无需重启即可成功看到输入nvidia-smi后熟悉的界面

---

好网站：

oldpan.me

whoiscaesarbao.com

http://manaai.cn

---

heatmap定位： http://b.edu.51cto.com/samsung/course/video?lessonId=237444 dsntnn: https://zhuanlan.zhihu.com/p/53057942

网络结构： https://baijiahao.baidu.com/s?id=1590362274035183205&wfr=spider&for=pc

---

yolo笔记：

• 池化层容易丢失低层次的特征
• 调整卷积核的步幅来调整下采样的程度
• 上采样能帮助网络学习细粒度的特征
• https://github.com/eriklindernoren/PyTorch-YOLOv3
• https://github.com/jwchoi384/Gaussian_YOLOv3
• https://github.com/ayooshkathuria/YOLO_v3_tutorial_from_scratch
• https://github.com/shaoshengsong/MobileNetV3-SSD

---

import time
import sys
import cv2
import numpy as np
import datetime
from PyQt5.QtCore import QObject, pyqtSignal, QThread, QTimer, Qt
from PyQt5.QtWidgets import (QProgressBar, QApplication, 
                            QPushButton, QDialog, 
                            QGridLayout, QMainWindow, 
                            QVBoxLayout, QWidget, 
                            QLabel)
from PyQt5.QtGui import QImage, QPixmap


srcHeight=480
srcWidth=480



class recoProcess(QObject):
    _signal = pyqtSignal(int)
    finished = pyqtSignal()

    def __init__(self):
        super(recoProcess, self).__init__()

    def run(self):
        for i in range(101):
            print(i)
            self._signal.emit(i)
            time.sleep(0.01)
        self.finished.emit()


class Video():
    def __init__(self, capture):
        self.capture = capture
        capture.set(3,srcWidth) # set Width
        capture.set(4,srcHeight) # set Height
        self.currentFrame = np.array([])
 
    def captureFrame(self):
        ret, readFrame = self.capture.read()
        return readFrame
 
    def captureNextFrame(self):
        ret, readFrame = self.capture.read()
        if (ret == True):
 
            # readFrame=cv2.resize(readFrame, (int(srcWidth / 4), int(srcHeight / 4)))
             
            #cv2.waitKey(1)
            self.currentFrame = cv2.cvtColor(readFrame, cv2.COLOR_BGR2RGB)
 
    def convertFrame(self):
        try:
            height, width = self.currentFrame.shape[:2]
            # print(height, width)
            img = QImage(self.currentFrame, width, height, QImage.Format_RGB888)
            img = QPixmap.fromImage(img)
            #self.previousFrame = self.currentFrame
            return img
        except:
            print('convertFrame error')
            return None





class CameraWin(QMainWindow):
    def __init__(self):
        super(CameraWin, self).__init__()

        self.init_ui()
 
    def init_ui(self):
        self.setWindowTitle('fingerReco')

        self.layout = QVBoxLayout()


        self.video = Video(cv2.VideoCapture(0))

        self.videoFrame = QLabel('VideoCapture')
        # self.videoFrame.setAlignment(Qt.AlignCenter)

        self.progress = QProgressBar(self)
        self.recoButton = QPushButton('识别', self)

        self.layout.addWidget(self.videoFrame)
        self.layout.addWidget(self.progress)
        self.layout.addWidget(self.recoButton)

        self.main_frame = QWidget()
        
        self.main_frame.setLayout(self.layout)

        self.recoButton.clicked.connect(self.recoButton_clicked)


        self._timer = QTimer(self)
        self._timer.timeout.connect(self.play)
        self._timer.start(2)
        self.update()

        
        self.setCentralWidget(self.main_frame)

        self.ret, self.capturedFrame = self.video.capture.read()

    def play(self):
        try:
            # nowTime=datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S')
            # print(nowTime)
            self.video.captureNextFrame()
            self.videoFrame.setPixmap(self.video.convertFrame())
            self.videoFrame.setScaledContents(True)
        except TypeError:
            print('No Frame')

    def recoButton_clicked(self):
        self.recoProcess_worker = recoProcess()
        self.recoProcessing_thread = QThread()
        self.recoProcess_worker.moveToThread(self.recoProcessing_thread)
        self.recoProcess_worker._signal.connect(self.progress.setValue)
        self.recoProcess_worker.finished.connect(self.recoProcess_worker_finished)
        self.recoProcessing_thread.started.connect(self.recoProcess_worker.run)
        self.recoProcessing_thread.start()

        self.recoButton.setEnabled(False)


    def recoProcess_worker_finished(self):
        self.recoProcessing_thread.quit()
        self.recoProcessing_thread.wait()
        self.recoButton.setEnabled(True)




if __name__ == '__main__':

    app = QApplication(sys.argv)
    win = CameraWin()
    win.show()
    app.exec_()

---

centernet c++

• https://blog.csdn.net/u011681952/article/details/92765549

---

sd卡烧录后恢复容量

1、管理员方式启动cmd

2、输入  diskpart

3、list  disk   #查看当前有哪些磁盘

4、select disk 1 #选择第一个磁盘 （这里我的sd卡是 disk 1）

5、clean

6、create partition primary #创建磁盘分区

7、active #激活磁盘分区

8、format fs=fat32 quick #已fat32格式快速格式化磁盘分区

---

jetson nano pytorch

• https://devtalk.nvidia.com/default/topic/1049071/jetson-nano/pytorch-for-jetson-nano/
• 扩展更多的swap分区避免Jetson Nano 4G的内存爆掉
• 如何把系统装在外置SSD中而不用忍受SD卡的速度
• running full yolov3 on the nano using either TensorRT or Deepstream 4.0.1?

Have you tried our deepstream SDK?
It contains the samples for YOLOv2, YOLOv2_tiny, YOLOv3 and YOLOv3_tiny model.

https://developer.nvidia.com/deepstream-sdk
/opt/nvidia/deepstream/deepstream-4.0/sources/objectDetector_Yolo

---

PID, line regression:

• https://openmv.io/blogs/news/linear-regression-line-following https://github.com/ivmech/ivPID https://book.openmv.cc/project/follow-lines.html

---

麦克纳姆轮 控制

https://zhuanlan.zhihu.com/p/20282234

import math


wheelJoints=[]
wheelJoints[1]='rollingJoint_fl'
wheelJoints[2]='rollingJoint_rl'
wheelJoints[3]='rollingJoint_rr'
wheelJoints[4]='rollingJoint_fr'

forwBackVelRange=[-240*math.pi/180,240*math.pi/180]
leftRightVelRange=[-240*math.pi/180,240*math.pi/180]
rotVelRange=[-240*math.pi/180,240*math.pi/180]

# (0~1000)
sliderForBack=600
sliderLeftRight=600
sliderRot=500

forwBackVel=forwBackVelRange[0]+sliderForBack*0.001*(forwBackVelRange[1]-forwBackVelRange[0])
leftRightVel=leftRightVelRange[0]+sliderLeftRight*0.001*(leftRightVelRange[1]-leftRightVelRange[0])
rotVel=rotVelRange[0]+sliderRot*0.001*(rotVelRange[1]-rotVelRange[0])

wheelJoints1V=-forwBackVel-leftRightVel-rotVel
wheelJoints2V=-forwBackVel+leftRightVel-rotVel
wheelJoints3V=-forwBackVel-leftRightVel+rotVel
wheelJoints4V=-forwBackVel+leftRightVel+rotVel

print(wheelJoints1V, wheelJoints2V, wheelJoints3V, wheelJoints4V)

error: command '/usr/bin/nvcc' failed with exit status 1 facebookresearch/maskrcnn-benchmark#25

荔枝派

https://licheezero.readthedocs.io/zh/latest/index.html

机器人仿真：

https://blog.csdn.net/weixin_39901416/article/details/78898266 http://www.coppeliarobotics.com/downloads.html https://blog.csdn.net/weixin_39059031/article/details/85017020 https://blog.csdn.net/DanielDingshengli/article/details/100640981

jetson nano 外接硬盘

https://zhuanlan.zhihu.com/p/65391640 https://www.liangzl.com/get-article-detail-134999.html