<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[LMengi000的博客]]></title><description><![CDATA[]]></description><link>https://blog.csdn.net/LMengi000</link><language>zh-cn</language><generator>https://blog.csdn.net/</generator><copyright><![CDATA[Copyright &copy; LMengi000]]></copyright><item><title><![CDATA[nvidia a100-pcie-40gb环境安装]]></title><link>https://blog.csdn.net/LMengi000/article/details/136403315</link><guid>https://blog.csdn.net/LMengi000/article/details/136403315</guid><author>LMengi000</author><pubDate>Fri, 01 Mar 2024 18:16:33 +0800</pubDate><description><![CDATA[1.conda create --name torch_li python=3.8
2.conda install pytorch1.7.1 torchvision0.8.2 torchaudio==0.7.2 cudatoolkit=11.0 -c pytorch
3.conda remove -n torch_li --all
4.pip install opencv-python-headless
5.pip install matplotlib
6.pip install imageio


]]></description><category></category></item><item><title><![CDATA[Linux下载、安装和配置anaconda]]></title><link>https://blog.csdn.net/LMengi000/article/details/135427179</link><guid>https://blog.csdn.net/LMengi000/article/details/135427179</guid><author>LMengi000</author><pubDate>Sat, 06 Jan 2024 15:32:52 +0800</pubDate><description><![CDATA[跟随这个链接下载安装anaconda：如何在Linux服务器上安装Anaconda（超详细） 链接:https://blog.csdn.net/wyf2017/article/details/118676765。安装sudo：链接：https://blog.csdn.net/hello_1995/article/details/109222650。安装screen：链接：https://blog.csdn.net/ztmajor/article/details/117651176。]]></description><category></category></item><item><title><![CDATA[配置anaconda+安装cuda和cudnn+将cuda和cudnn配置到指定目录+使用cuda命令安装cuda和cudnn]]></title><link>https://blog.csdn.net/LMengi000/article/details/133992724</link><guid>https://blog.csdn.net/LMengi000/article/details/133992724</guid><author>LMengi000</author><pubDate>Mon, 23 Oct 2023 16:19:48 +0800</pubDate><description><![CDATA[如果在安装过程中出现其他问题，就根据问题进行百度搜索。]]></description><category></category></item><item><title><![CDATA[ref_affine矩阵含义]]></title><link>https://blog.csdn.net/LMengi000/article/details/118155052</link><guid>https://blog.csdn.net/LMengi000/article/details/118155052</guid><author>LMengi000</author><pubDate>Wed, 23 Jun 2021 15:52:18 +0800</pubDate><description><![CDATA[图片传送门-------------&gt;传送门

]]></description><category></category></item><item><title><![CDATA[注意力机制（未完待续）]]></title><link>https://blog.csdn.net/LMengi000/article/details/117334156</link><guid>https://blog.csdn.net/LMengi000/article/details/117334156</guid><author>LMengi000</author><pubDate>Fri, 28 May 2021 14:25:57 +0800</pubDate><description><![CDATA[实现常见CNN网络结构中添加注意力(attention)机制 - 简书 (jianshu.com)

(295条消息) 注意力机制代码_pytorch中加入注意力机制（CBAM），以ResNet为例_weixin_39787089的博客-CSDN博客

大话注意力机制（Attention Mechanism） - 雪饼的个人空间 - OSCHINA - 中文开源技术交流社区

GitHub - bubbliiiing/Keras-Attention: 这个仓库主要包含了LSTM、卷积神经网络中，注意力机制的]]></description><category></category></item><item><title><![CDATA[报错：AttributeError: ‘str‘ object has no attribute ‘decode‘]]></title><link>https://blog.csdn.net/LMengi000/article/details/117306270</link><guid>https://blog.csdn.net/LMengi000/article/details/117306270</guid><author>LMengi000</author><pubDate>Wed, 26 May 2021 21:42:02 +0800</pubDate><description><![CDATA[参考这篇博客，更新hdf5
]]></description><category></category></item><item><title><![CDATA[TensorFlow - GAN实现训练和测试，mnist数据集]]></title><link>https://blog.csdn.net/LMengi000/article/details/117305722</link><guid>https://blog.csdn.net/LMengi000/article/details/117305722</guid><author>LMengi000</author><pubDate>Wed, 26 May 2021 21:30:30 +0800</pubDate><description><![CDATA[遇到的问题： 

我在代码：generator.load_weights('D:/imageRegistration/languageGan/MyGAN/models/G_model/450.h5',by_name = True) 中加入了by_name=True 这一句，测试的过程中，测试生成器时，一直生成的还是噪声图，但是在输出的图片文件夹中，输出的图片中数字的效果还是很好的，我把这一句去掉之后，测试生成器，生成的图片效果就与图片中的效果差不多。



keras源码engine中t...]]></description><category></category></item><item><title><![CDATA[TensorFlow-用cifar10数据集搭建神经网络]]></title><link>https://blog.csdn.net/LMengi000/article/details/117194555</link><guid>https://blog.csdn.net/LMengi000/article/details/117194555</guid><author>LMengi000</author><pubDate>Sun, 23 May 2021 14:59:35 +0800</pubDate><description><![CDATA[# -*- coding: utf-8 -*-
"""
Created on Sun May 23 14:29:45 2021

@author: Lenovo
"""
import tensorflow as tf
import os
import numpy as np
from matplotlib import pyplot as plt
from tensorflow.keras.layers import Conv2D, BatchNormalization, Activation, Ma...]]></description><category></category></item><item><title><![CDATA[TensorFlow-Cifar10数据集]]></title><link>https://blog.csdn.net/LMengi000/article/details/117194193</link><guid>https://blog.csdn.net/LMengi000/article/details/117194193</guid><author>LMengi000</author><pubDate>Sun, 23 May 2021 13:39:07 +0800</pubDate><description><![CDATA[# -*- coding: utf-8 -*-
"""
Created on Sun May 23 13:24:43 2021

@author: LiMeng
"""
import tensorflow as tf
from matplotlib import pyplot as plt
import numpy as np

np.set_printoptions(threshold = np.inf)
cifar10 = tf.keras.datasets.cifar10
(x_train,...]]></description><category></category></item><item><title><![CDATA[TensorFlow-描述卷积计算层、批标准化、池化、舍弃（Dropout）]]></title><link>https://blog.csdn.net/LMengi000/article/details/117193463</link><guid>https://blog.csdn.net/LMengi000/article/details/117193463</guid><author>LMengi000</author><pubDate>Sun, 23 May 2021 13:05:20 +0800</pubDate><description><![CDATA[卷积层



批标准化







池化





舍弃（Dropout）


 为了缓解神经网络过拟合，常把隐藏层的部分神经元按照一定比例从神经网络中临时舍弃。在使用神经网络时，再把所有神经元回复到神经网络中。




总结


卷积就是特征提取器。





...]]></description><category></category></item><item><title><![CDATA[TensorFlow-全零填充]]></title><link>https://blog.csdn.net/LMengi000/article/details/117193353</link><guid>https://blog.csdn.net/LMengi000/article/details/117193353</guid><author>LMengi000</author><pubDate>Sun, 23 May 2021 12:37:19 +0800</pubDate><description><![CDATA[在吴恩达卷积神经网络这个视频中，我之前听过一次，吴恩达的视频里面对公式的讲解也很详细，可以参考一下吴恩达的卷积神经网络的视频。



]]></description><category></category></item><item><title><![CDATA[TensroFlow-感受野]]></title><link>https://blog.csdn.net/LMengi000/article/details/117192782</link><guid>https://blog.csdn.net/LMengi000/article/details/117192782</guid><author>LMengi000</author><pubDate>Sun, 23 May 2021 12:33:27 +0800</pubDate><description><![CDATA[感受野：是指输出特征图中1个像素点，映射到原始输入图片的区域大小，





如果对这个5*5的原始输入图片，用黄色的3*3的卷积核作用，会输出一个3*3的输出特征图。

这个3*3的特征图上的每一个点，映射到原始5*5的图片中，是一个3*3的区域，所以它的感受野是3。





如果再对这个3*3的特征图，再用一个3*3的卷积核作用一下，就会输出一个1*1的输出特征图。

这个1*1的特征图上的像素点，映射到原始图片是5*5的区域，所以它的感受野是5. 。





如果对原始的5*5的图片，直接...]]></description><category></category></item><item><title><![CDATA[TensorFlow-卷积计算的过程]]></title><link>https://blog.csdn.net/LMengi000/article/details/117190264</link><guid>https://blog.csdn.net/LMengi000/article/details/117190264</guid><author>LMengi000</author><pubDate>Sun, 23 May 2021 11:11:18 +0800</pubDate><description><![CDATA[如果你觉得某层特征提取的能力不足，可以在这一层多用几个卷积核提高这一层的特征提取能力。









]]></description><category></category></item><item><title><![CDATA[最原始的GAN-我开始慢慢的懂GAN了]]></title><link>https://blog.csdn.net/LMengi000/article/details/117168791</link><guid>https://blog.csdn.net/LMengi000/article/details/117168791</guid><author>LMengi000</author><pubDate>Sat, 22 May 2021 18:52:47 +0800</pubDate><description><![CDATA[参考一篇非常全面的博文（传送门），把最原始的GAN网络过了一遍，虽然还是心中有疑虑，还是有问题，但是好事多磨，总会想明白的。

自己想不懂的问题：

1.在训练网络（generator）中，它是怎么生成一个虚假图片的？

build_generator（）这个函数就是一个全连接网络，都说generator就是来生成一个虚假图片的，但是最后build_generator返回不是一个造好的虚假图片啊！

在代码部分：

 noise= np.random.normal(0,1, (batch_size, 1..]]></description><category></category></item><item><title><![CDATA[在Anaconda搭配VoxelMorph的环境]]></title><link>https://blog.csdn.net/LMengi000/article/details/117062471</link><guid>https://blog.csdn.net/LMengi000/article/details/117062471</guid><author>LMengi000</author><pubDate>Thu, 20 May 2021 09:19:21 +0800</pubDate><description><![CDATA[起初，搭配环境是一件繁琐的事情，一直在anaconda中装了卸载，卸载了再安装，看了网上很多博文，最终选择一篇介绍搭配VoxelMorph环境的介绍，觉得写的很好，值得收藏。

【VoxelMorph论文共读】002 代码环境准备

...]]></description><category></category></item><item><title><![CDATA[TensorFlow-给图识物]]></title><link>https://blog.csdn.net/LMengi000/article/details/116737578</link><guid>https://blog.csdn.net/LMengi000/article/details/116737578</guid><author>LMengi000</author><pubDate>Thu, 13 May 2021 10:02:49 +0800</pubDate><description><![CDATA[from PIL import Image
import numpy as np
import tensorflow as tf
model_save_path = './checkpoint/mnist.ckpt'

model = tf.keras.models.Sequential([
    tf.keras.layers.Flatten(),
    tf.keras.layers.Dense(128,activation='relu'),
    tf.keras.layers.D...]]></description><category></category></item><item><title><![CDATA[TensorFlow——acc/loss 可视化]]></title><link>https://blog.csdn.net/LMengi000/article/details/116736808</link><guid>https://blog.csdn.net/LMengi000/article/details/116736808</guid><author>LMengi000</author><pubDate>Thu, 13 May 2021 09:42:17 +0800</pubDate><description><![CDATA[import tensorflow as tf
import os
#导入numpy模块
import numpy as np

from matplotlib import pyplot as plt

np.set_printoptions(threshold=np.inf)

mnist = tf.keras.datasets.mnist
(x_train,y_train),(x_test,y_test) = mnist.load_data()
x_trian,x_test = x_tr...]]></description><category></category></item><item><title><![CDATA[TensorFlow-参数提取]]></title><link>https://blog.csdn.net/LMengi000/article/details/116736399</link><guid>https://blog.csdn.net/LMengi000/article/details/116736399</guid><author>LMengi000</author><pubDate>Thu, 13 May 2021 09:26:19 +0800</pubDate><description><![CDATA[import tensorflow as tf
import os
#导入numpy模块
import numpy as np
np.set_printoptions(threshold=np.inf)

mnist = tf.keras.datasets.mnist
(x_train,y_train),(x_test,y_test) = mnist.load_data()
x_trian,x_test = x_train / 255.0,x_test / 255.0

model = tf...]]></description><category></category></item><item><title><![CDATA[TensorFlow-断点续训]]></title><link>https://blog.csdn.net/LMengi000/article/details/116735020</link><guid>https://blog.csdn.net/LMengi000/article/details/116735020</guid><author>LMengi000</author><pubDate>Thu, 13 May 2021 09:18:25 +0800</pubDate><description><![CDATA[import tensorflow as tf
import os

mnist = tf.keras.datasets.mnist
(x_train,y_train),(x_test,y_test) = mnist.load_data()
x_trian,x_test = x_train / 255.0,x_test / 255.0

model = tf.keras.models.Sequential([
    tf.keras.layers.Flatten(),
    tf.keras....]]></description><category></category></item><item><title><![CDATA[TensorFlow-数据增强（扩充数据集）]]></title><link>https://blog.csdn.net/LMengi000/article/details/116565908</link><guid>https://blog.csdn.net/LMengi000/article/details/116565908</guid><author>LMengi000</author><pubDate>Sun, 09 May 2021 20:41:04 +0800</pubDate><description><![CDATA[图像增强就是对图像的简单形变

image_gen_train.fit(x_train)这里的fit需要输入一个思维的数据，所以要对x_train进行reshape，把60000张28行28列的数据，转换为60000张28行28列的单通道数据（即为1），单通道为灰度值。





import tensorflow as tf
from matplotlib import pyplot as plt
from tensorflow.keras.preprocessing.image impo...]]></description><category></category></item></channel></rss>