<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[Jad_Goh的博客]]></title><description><![CDATA[]]></description><link>https://blog.csdn.net/Jad_Goh</link><language>zh-cn</language><generator>https://blog.csdn.net/</generator><copyright><![CDATA[Copyright &copy; Jad_Goh]]></copyright><item><title><![CDATA[下载老版本vscode插件的方法]]></title><link>https://blog.csdn.net/Jad_Goh/article/details/135484574</link><guid>https://blog.csdn.net/Jad_Goh/article/details/135484574</guid><author>Jad_Goh</author><pubDate>Tue, 09 Jan 2024 16:44:17 +0800</pubDate><description><![CDATA[老版本vscode插件的下载方法]]></description><category></category></item><item><title><![CDATA[TT100K/BDD100K数据集格式转换]]></title><link>https://blog.csdn.net/Jad_Goh/article/details/129033999</link><guid>https://blog.csdn.net/Jad_Goh/article/details/129033999</guid><author>Jad_Goh</author><pubDate>Tue, 14 Feb 2023 21:59:57 +0800</pubDate><description><![CDATA[TT100K数据集，BDD100K数据集格式转换]]></description><category></category></item><item><title><![CDATA[HybridNets: End-to-End Perception Network]]></title><link>https://blog.csdn.net/Jad_Goh/article/details/127727768</link><guid>https://blog.csdn.net/Jad_Goh/article/details/127727768</guid><author>Jad_Goh</author><pubDate>Mon, 07 Nov 2022 13:47:42 +0800</pubDate><description><![CDATA[HybridNets的重要内容翻译。其主要改进点为：主干网络替换为EfficientNet-B3；颈部网络替换为BiFPN；分割损失修改为Focal损失+Tversky损失；后者可以改善类别不平衡问题。]]></description><category></category></item><item><title><![CDATA[YOLOPv2: Better, Faster, Stronger for Panoptic Driving Perception（2022）]]></title><link>https://blog.csdn.net/Jad_Goh/article/details/127704342</link><guid>https://blog.csdn.net/Jad_Goh/article/details/127704342</guid><author>Jad_Goh</author><pubDate>Sat, 05 Nov 2022 21:27:52 +0800</pubDate><description><![CDATA[YOLOPv2相较于YOLOP改动较少，主干网络改为E-ELAN、可行驶区域分割分支接在主干网络的末端、车道线损失函数改为Focal Loss+Dice Loss和采用了反卷积。]]></description><category></category></item><item><title><![CDATA[You Only Look at Once for Panoptic driving Perception（2021）]]></title><link>https://blog.csdn.net/Jad_Goh/article/details/127677907</link><guid>https://blog.csdn.net/Jad_Goh/article/details/127677907</guid><author>Jad_Goh</author><pubDate>Sat, 05 Nov 2022 13:36:54 +0800</pubDate><description><![CDATA[YOLOP人工翻译，基本为全文翻译和重新排版，使得正文和图表关联更紧密已读，仅有部分不重要内容进行了省略。]]></description><category></category></item><item><title><![CDATA[orin环境配置问题记录]]></title><link>https://blog.csdn.net/Jad_Goh/article/details/126849652</link><guid>https://blog.csdn.net/Jad_Goh/article/details/126849652</guid><author>Jad_Goh</author><pubDate>Wed, 14 Sep 2022 16:08:07 +0800</pubDate><description><![CDATA[orin配置问题记录]]></description><category></category></item><item><title><![CDATA[深度学习中的激活函数]]></title><link>https://blog.csdn.net/Jad_Goh/article/details/126425904</link><guid>https://blog.csdn.net/Jad_Goh/article/details/126425904</guid><author>Jad_Goh</author><pubDate>Fri, 19 Aug 2022 16:56:07 +0800</pubDate><description><![CDATA[深度学习中常用的激活函数优缺点分析：sigmoid、tanh、ReLU...]]></description><category></category></item><item><title><![CDATA[yolov5中两处scale讨论]]></title><link>https://blog.csdn.net/Jad_Goh/article/details/124533439</link><guid>https://blog.csdn.net/Jad_Goh/article/details/124533439</guid><author>Jad_Goh</author><pubDate>Sun, 01 May 2022 23:31:46 +0800</pubDate><description><![CDATA[第一处：hyp文件中的scale: 0.5；

第二处：train文件中multi-scale（if opt.multi_scale: sz = random.randrange(imgsz * 0.5, imgsz * 1.5 + gs) // gs * gs # size）。

结论1:hyp.scale = 0.5(假如input=640)就代表着参与训练的图片的输入可能是320~640×1.5(小于640的部分会被补上黑边，大于640的部分会被裁掉)。

结论2:train的时候选择--multi-]]></description><category></category></item><item><title><![CDATA[jupyter notebook 相关配置]]></title><link>https://blog.csdn.net/Jad_Goh/article/details/122226320</link><guid>https://blog.csdn.net/Jad_Goh/article/details/122226320</guid><author>Jad_Goh</author><pubDate>Wed, 29 Dec 2021 22:49:38 +0800</pubDate><description><![CDATA[1.安装插件nbextensions


pip install jupyter_contrib_nbextensions &amp;&amp; jupyter contrib nbextension install

下面4个常用插件：

1）Table of Contents：更容易导航

2）Autopep8：轻轻一击就能获得简洁代码，使代码符合格式

3）variable inspector

4）ExecuteTime

2.添加新环境到内核


conda activate mark  # 激活]]></description><category></category></item><item><title><![CDATA[Ubuntu 常见问题汇总]]></title><link>https://blog.csdn.net/Jad_Goh/article/details/122193485</link><guid>https://blog.csdn.net/Jad_Goh/article/details/122193485</guid><author>Jad_Goh</author><pubDate>Tue, 28 Dec 2021 15:16:15 +0800</pubDate><description><![CDATA[1.双系统网络连接问题

这个Lyon博主已经说的很详细了（链接1）（链接2），这边做一个备份。


# 1.启动NetworkManager
sudo vim /etc/NetworkManager/NetworkManager.conf #修改managed=true
# 2.重启服务
sudo service network-manager restart
# 查看启用的网卡
ifconfig
# 查看所有网卡
ifconfig -a
# 3.如果nmcli命令查看网卡状态，发现网络没有被networ]]></description><category></category></item><item><title><![CDATA[解析rosbag中的.bag文件，得到.jpg图片数据和.pcd点云数据]]></title><link>https://blog.csdn.net/Jad_Goh/article/details/121771722</link><guid>https://blog.csdn.net/Jad_Goh/article/details/121771722</guid><author>Jad_Goh</author><pubDate>Tue, 07 Dec 2021 16:29:59 +0800</pubDate><description><![CDATA[一.查看.bag文件中的数据信息


rosbag info *.bag

结果如下所示：



二.解析.bag文件到.jpg文件

方法一（存在解析出的图片数量与msg数量不符的情况，方法二可解决）

1.使用touch命令建立export.launch文件, 文件内容如下所示，注意更改相应的.bag文件路径和topics名称。


touch export.launch


&lt;launch&gt;
      &lt;node pkg="rosbag" type="play" name="ro.]]></description><category></category></item><item><title><![CDATA[ROS Melodic配置记录]]></title><link>https://blog.csdn.net/Jad_Goh/article/details/121606836</link><guid>https://blog.csdn.net/Jad_Goh/article/details/121606836</guid><author>Jad_Goh</author><pubDate>Mon, 29 Nov 2021 13:13:09 +0800</pubDate><description><![CDATA[目录

一、参考官网进行安装：官网地址。

二、问题1：依赖报错问题。参照博客解决

三、问题2：rosdep update超时问题。参照博客解决。

一、参考官网进行安装：官网地址。

二、问题1：依赖报错问题。参照博客解决


sudo apt install ros-melodic-desktop-full #依赖报错
sudo aptitude install ros-melodic-desktop ros-melodic-perception ros-melodic-simulators ros-]]></description><category></category></item><item><title><![CDATA[second.pytorch 1.6&1.5.1配置实录]]></title><link>https://blog.csdn.net/Jad_Goh/article/details/117400523</link><guid>https://blog.csdn.net/Jad_Goh/article/details/117400523</guid><author>Jad_Goh</author><pubDate>Sun, 30 May 2021 22:44:33 +0800</pubDate><description><![CDATA[参考：traveller59/second.pytorch

环境：


]]></description><category></category></item><item><title><![CDATA[重装Ubuntu18.04步骤]]></title><link>https://blog.csdn.net/Jad_Goh/article/details/117395142</link><guid>https://blog.csdn.net/Jad_Goh/article/details/117395142</guid><author>Jad_Goh</author><pubDate>Sun, 30 May 2021 14:48:09 +0800</pubDate><description><![CDATA[清空磁盘：我用的是DiskGenius
	制作启动盘：我用的是EasyBCD2.4
	进入Boot界面选择Boot优先级，禁用
]]></description><category></category></item></channel></rss>