<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[SH_13的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/B_DATA_NUIST</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; B_DATA_NUIST]]></copyright><item><title><![CDATA[2021-09-08]]></title><link>https://blog.youkuaiyun.com/B_DATA_NUIST/article/details/120178110</link><guid>https://blog.youkuaiyun.com/B_DATA_NUIST/article/details/120178110</guid><author>B_DATA_NUIST</author><pubDate>Wed, 08 Sep 2021 14:42:13 +0800</pubDate><description><![CDATA[最远点采样（Farest Point Sampling）
import numpy as np
import matplotlib.pyplot as plt

class FarthestSample(object):
    def __init__(self):
        super(FarthestSample, self).__init__()
    
    def __cal_distances(self, p_0, points):
        return ((p_0 -  p]]></description><category></category></item><item><title><![CDATA[设置CUDA_VISIBLE_DEVICES的方法]]></title><link>https://blog.youkuaiyun.com/B_DATA_NUIST/article/details/107973053</link><guid>https://blog.youkuaiyun.com/B_DATA_NUIST/article/details/107973053</guid><author>B_DATA_NUIST</author><pubDate>Tue, 15 Sep 2020 00:13:14 +0800</pubDate><description><![CDATA[通过bash脚本设置CUDA_VISIBLE_DEVICES
set CUDA_VISIBLE_DEVICES=0, 1, 2, 3
source ~/anaconda3/bin/activate torch
python -c 'import torch; print(torch.__version__); print(torch.cuda.device_count)'



]]></description><category></category></item><item><title><![CDATA[torch.nn.Module中modules和children的区别]]></title><link>https://blog.youkuaiyun.com/B_DATA_NUIST/article/details/108588678</link><guid>https://blog.youkuaiyun.com/B_DATA_NUIST/article/details/108588678</guid><author>B_DATA_NUIST</author><pubDate>Tue, 15 Sep 2020 00:04:42 +0800</pubDate><description><![CDATA[torch.nn.Module中modules()和children()的区别
pytorch中一切网络结构一般都是继承自nn.Module类来实现的。nn.Module中存在两个相似的method：torch.nn.modules()和torch.nn.Modules.children()

]]></description><category></category></item><item><title><![CDATA[python-pcl计算点云法向量]]></title><link>https://blog.youkuaiyun.com/B_DATA_NUIST/article/details/106104081</link><guid>https://blog.youkuaiyun.com/B_DATA_NUIST/article/details/106104081</guid><author>B_DATA_NUIST</author><pubDate>Mon, 14 Sep 2020 23:42:05 +0800</pubDate><description><![CDATA[import pcl
import pcl.pcl_visualization

def main():
	cloud = pcl.load('filename.pcd') # load the pcd file
	ne = cloud.make_NormalEstimation() # instance the NormalEstimation 
	ne.set_KSearch(20) # set the KNN search range of NormalEstimation
	normals = ne]]></description><category></category></item><item><title><![CDATA[如何设置随机种子保证pytorch代码的可复现性]]></title><link>https://blog.youkuaiyun.com/B_DATA_NUIST/article/details/105857043</link><guid>https://blog.youkuaiyun.com/B_DATA_NUIST/article/details/105857043</guid><author>B_DATA_NUIST</author><pubDate>Thu, 30 Apr 2020 15:52:53 +0800</pubDate><description><![CDATA[在深度学习研究领域,论文结果的可复现性是一个很大的问题.遑论各种paper中的代码,有时候就是自己写的代码,都难以保证可复现性:即使使用同样的网络结构,同样的数据库,在同一台机器上训练,训练的结果都有差别.这一现象很大程度上是由于深度学习训练过程中的随机性造成的.

网络参数的随机初始化
正则化方法,例如dropout在训练中随机丢弃网络中的节点
优化过程,例如SGD,RMSPorp或者Adam等...]]></description><category></category></item><item><title><![CDATA[将rgb颜色编码成一个float32浮点数]]></title><link>https://blog.youkuaiyun.com/B_DATA_NUIST/article/details/105835142</link><guid>https://blog.youkuaiyun.com/B_DATA_NUIST/article/details/105835142</guid><author>B_DATA_NUIST</author><pubDate>Wed, 29 Apr 2020 12:08:07 +0800</pubDate><description><![CDATA[使用python的pypcd包读写pcd文件,需要将[r, g, b]三通道颜色转换成一个float32数和与之对应的逆变换,分别称作编码器和解码器
def encode_rgb_for_pcl(rgb):
    """ Encode bit-packed RGB for use with PCL.
    :param rgb: Nx3 uint8 array with RGB values....]]></description><category></category></item><item><title><![CDATA[Numpy中arctan和arctan2的区别]]></title><link>https://blog.youkuaiyun.com/B_DATA_NUIST/article/details/105792308</link><guid>https://blog.youkuaiyun.com/B_DATA_NUIST/article/details/105792308</guid><author>B_DATA_NUIST</author><pubDate>Mon, 27 Apr 2020 19:25:13 +0800</pubDate><description><![CDATA[Numpy中arctan和arctan2的区别缘起功能快捷键合理的创建标题，有助于目录的生成如何改变文本的样式插入链接与图片如何插入一段漂亮的代码片生成一个适合你的列表创建一个表格设定内容居中、居左、居右SmartyPants创建一个自定义列表如何创建一个注脚注释也是必不可少的KaTeX数学公式新的甘特图功能，丰富你的文章UML 图表FLowchart流程图导出与导入导出导入
缘起
在工作中需要将...]]></description><category></category></item></channel></rss>