<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[yeahDeDiQiZhang的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; yeahDeDiQiZhang]]></copyright><item><title><![CDATA[Keras实例：估计模型训练所需样本量]]></title><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/117691331</link><guid>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/117691331</guid><author>yeahDeDiQiZhang</author><pubDate>Tue, 08 Jun 2021 10:19:15 +0800</pubDate><description><![CDATA[这是一篇来自 Keras 官方文档中的案例
作者的想法很简单，拿一个小数据集：
dataset_name = "tf_flowers"
batch_size = 64
image_size = (224, 224)

90% 用于训练，10% 用于预测
(train_data, test_data), ds_info = tfds.load(
    dataset_name,
    split=["train[:90%]", "train[90%:]"],
    shuffle_files=True,]]></description><category></category></item><item><title><![CDATA[Python内存管理指南]]></title><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/101705039</link><guid>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/101705039</guid><author>yeahDeDiQiZhang</author><pubDate>Mon, 30 Sep 2019 10:41:56 +0800</pubDate><description><![CDATA[Python内存管理指南
原文地址：https://towardsdatascience.com/memory-management-in-python-6bea0c8aecc9
了解内存管理对于软件开发人员非常重要。 随着Python在软件开发中广泛的使用，编写高效的Python代码通常意味着编写内存高效的代码。 随着大数据的使用越来越多，内存管理的重要性不容忽视。 无效的内存管理会导致应用程序...]]></description><category></category></item><item><title><![CDATA[Neo4j - springboot 开发中的坑]]></title><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/101294959</link><guid>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/101294959</guid><author>yeahDeDiQiZhang</author><pubDate>Tue, 24 Sep 2019 17:17:36 +0800</pubDate><description><![CDATA[Neo4j java开发中的坑
最近给 Neo4j 写调用接口，遇到一个大坑花了一周多时间，记录一下
项目内容是给定一个节点，从 Neo4j 查询该节点对应的所有关系与节点。
如
在这里参考了很多文章，例举如下

Spring boot项目集成Neo4j
Spring-Boot+Neo4j+节点之间关系网的搭建和查询

结果定义了节点后，只能查出  (n:药品名称)-&gt;() 的内容，(n:药...]]></description><category></category></item><item><title><![CDATA[分辨合格数据科学家的五道面试题]]></title><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/83542364</link><guid>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/83542364</guid><author>yeahDeDiQiZhang</author><pubDate>Tue, 30 Oct 2018 17:00:24 +0800</pubDate><description><![CDATA[Five Interview Questions to Predict a Good Data Scientist
原文地址：https://medium.com/predict/five-interview-questions-to-predict-a-good-data-scientist-40d310cdcd68
在我们这个行业，经常会提到数据科学家严重短缺这个问题。这种问题会造成如机器学习...]]></description><category></category></item><item><title><![CDATA[Mac下使用dlib的imglab标注图片]]></title><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/83346480</link><guid>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/83346480</guid><author>yeahDeDiQiZhang</author><pubDate>Wed, 24 Oct 2018 16:41:50 +0800</pubDate><description><![CDATA[官方文档见 https://github.com/davisking/dlib/tree/master/tools/imglab
安装依赖
需要先安装 cmake 和 X11，
cmake 通过下面命令安装
$ brew install cmake

X11 需要下载安装 https://www.xquartz.org/ ，直接安装后，需要创建软连接，命令如下
$ cd /usr/local/op...]]></description><category></category></item><item><title><![CDATA[Cascaded Pyramid Network for Multi-Person Pose Estimation]]></title><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/83061284</link><guid>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/83061284</guid><author>yeahDeDiQiZhang</author><pubDate>Tue, 16 Oct 2018 11:48:20 +0800</pubDate><description><![CDATA[Cascaded Pyramid Network是2017年 COCO 人体姿态点检测时提出的网络，并获得了冠军。]]></description><category></category></item><item><title><![CDATA[雪浪制造AI挑战赛—视觉计算辅助良品检测]]></title><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/82385420</link><guid>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/82385420</guid><author>yeahDeDiQiZhang</author><pubDate>Tue, 04 Sep 2018 17:09:10 +0800</pubDate><description><![CDATA[比赛网址：https://tianchi.aliyun.com/competition/introduction.htm?spm=5176.100150.711.5.2f5f2784nAcZxN&amp;amp;raceId=231666



这个比赛的任务是分类，初赛和复赛规则不同，初赛为二分类，而复赛为十一分类。

观察图像，目标为分类图像中布料上瑕疵的种类，而瑕疵大小不一，有的很长，有的很小，只有...]]></description><category></category></item><item><title><![CDATA[[kaggle]Humpback Whale Identification Challenge冠军方案]]></title><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/81450724</link><guid>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/81450724</guid><author>yeahDeDiQiZhang</author><pubDate>Wed, 08 Aug 2018 17:58:05 +0800</pubDate><description><![CDATA[Whale Recognition Model with score 0.78563



原文地址：https://www.kaggle.com/martinpiotte/whale-recognition-model-with-score-0-78563



蓝色字体为译者注



鲸鱼分类模型

本文记述了座头鲸挑战赛0.78563提交成绩的算法策略。 
本文应该与 Bounding Bo...]]></description><category></category></item><item><title><![CDATA[FashionAI全球挑战赛—服饰关键点定位]]></title><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/80621258</link><guid>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/80621258</guid><author>yeahDeDiQiZhang</author><pubDate>Fri, 08 Jun 2018 11:53:24 +0800</pubDate><description><![CDATA[比赛官网地址：https://tianchi.aliyun.com/competition/introduction.htm?spm=5176.100150.711.5.5f902784pJIfNg&amp;amp;amp;amp;amp;amp;amp;raceId=231648



在初赛时使用的Hourglass网络，在初赛第二轮的时候机器的水泵烧了。。到复赛快结束才修好。。所以复赛基本没有做改进。

网络结构参考我之前写的文章《Sta...]]></description><category></category></item><item><title><![CDATA[在Mac上运行TensorFlow.js的PoseNet demo]]></title><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/80281441</link><guid>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/80281441</guid><author>yeahDeDiQiZhang</author><pubDate>Fri, 11 May 2018 14:49:53 +0800</pubDate><description><![CDATA[项目地址：https://github.com/tensorflow/tfjs-models 

首先将项目clone下来


  git clone https://github.com/tensorflow/tfjs-models.git


进入PoseNet目录


  cd posenet/


使用yarn安装，如果电脑上没有安装yarn，可以先使用homebrew安装yarn


 ...]]></description><category></category></item><item><title><![CDATA[基于CNN的图形分类综合设计指南]]></title><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/80236786</link><guid>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/80236786</guid><author>yeahDeDiQiZhang</author><pubDate>Tue, 08 May 2018 17:29:31 +0800</pubDate><description><![CDATA[A Comprehensive Design Guide for Image Classification CNNs原文地址：https://hackernoon.com/a-comprehensive-design-guide-for-image-classification-cnns-46091260fb92你想做一个图像分类任务，但不知道从哪里开始。你该使用哪个预训练网络？ 你应该如何修改它...]]></description><category></category></item><item><title><![CDATA[Tensorflow使用教训：被tf.image.resize浪费的60天]]></title><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/79550113</link><guid>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/79550113</guid><author>yeahDeDiQiZhang</author><pubDate>Wed, 14 Mar 2018 11:21:55 +0800</pubDate><description><![CDATA[How Tensorflow’s tf.image.resize stole 60 days of my life原文地址：https://hackernoon.com/how-tensorflows-tf-image-resize-stole-60-days-of-my-life-aba5eb093f35?gi=f6c680bed9e0这是一个对所有计算机视觉工作者的简短警告：请勿使用任何tf....]]></description><category></category></item><item><title><![CDATA[[Kaggle]Plant Seedlings Classification]]></title><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/79541040</link><guid>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/79541040</guid><author>yeahDeDiQiZhang</author><pubDate>Tue, 13 Mar 2018 15:39:43 +0800</pubDate><description><![CDATA[Kaggle 上的比赛， 识别12种类型的植物，部分图片如下：预处理这里参考Gábor Vecsei的kernel，将叶片从背景中提取出来训练使用keras中的Xception模型，数据增加部分使用了[-180°，180°]的随机旋转，0.3的随机平移，0.3的随机放缩与随机翻转，代码如下：batch_size = 16
train_img = np.zeros([4750, img_size, ...]]></description><category></category></item><item><title><![CDATA[[Kaggle]Dog Breed Identification]]></title><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/79354824</link><guid>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/79354824</guid><author>yeahDeDiQiZhang</author><pubDate>Thu, 01 Mar 2018 14:54:22 +0800</pubDate><description><![CDATA[Kaggle 上的比赛，在这里使用了额外数据集Stanford Dogs Dataset。在Public Leaderboard上排名57/1286（top 5%)首先使用TensorFlow model zoo中的预训练模型导出图片特征，这里使用了inception_v3 与 resnet_v1_152import sys
sys.path.append('slim')
from tqdm im...]]></description><category></category></item><item><title><![CDATA[用TensorFlow lite将MobileNet移植到Android设备上]]></title><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/78925036</link><guid>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/78925036</guid><author>yeahDeDiQiZhang</author><pubDate>Thu, 28 Dec 2017 18:14:47 +0800</pubDate><description><![CDATA[分类目标是输入一张自行车图片，判断是山地车还是公路车。
第一步：
在百度图片分别爬取5000张山地车和公路车的图片，放于data/mountain和data/road两个文件夹下
第二步：
用TensorFlow自带的工具来fine-tuning训练mobilenet：
git clone https://github.com/tensorflow/tensorflow.git
cd t]]></description><category></category></item><item><title><![CDATA[Stacked Hourglass Networks]]></title><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/78578153</link><guid>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/78578153</guid><author>yeahDeDiQiZhang</author><pubDate>Sun, 19 Nov 2017 23:04:37 +0800</pubDate><description><![CDATA[http://www-personal.umich.edu/~alnewell/pose/]]></description><category></category></item><item><title><![CDATA[TensorFlow实现中文字体分类]]></title><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/78523182</link><guid>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/78523182</guid><author>yeahDeDiQiZhang</author><pubDate>Mon, 13 Nov 2017 18:19:49 +0800</pubDate><description><![CDATA[1.预处理
首先在网上找一份常用汉字大全，我这里找了一份2994字的常用汉字作为训练，712字的次常用汉字作为测试。
操作系统内就自带字体文件，后缀为ttc和ttf，Mac的路径为 /System/Library/Fonts，选取若干个作为分类的对象。
接着用PIL库来生成字体图片，生成的时候本想每个字居中显示。但是当一个字体中的字居中了，另外字体的字就会跑偏，因此这里用numpy来框出字体]]></description><category></category></item><item><title><![CDATA[python+selenium+chrome登录人人网爬取公共主页内容]]></title><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/78465080</link><guid>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/78465080</guid><author>yeahDeDiQiZhang</author><pubDate>Tue, 07 Nov 2017 10:23:51 +0800</pubDate><description><![CDATA[首先，创建一个driver


#!/usr/bin/env python
# -*- coding: utf-8 -*-

from selenium import webdriver

driver = webdriver.Chrome()
driver.get('http://www.renren.com/')



打开人人网首页，检查元素，分别获取email和password]]></description><category></category></item><item><title><![CDATA[Convolutional Pose Machines]]></title><link>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/78131566</link><guid>https://blog.youkuaiyun.com/yeahDeDiQiZhang/article/details/78131566</guid><author>yeahDeDiQiZhang</author><pubDate>Fri, 29 Sep 2017 10:39:15 +0800</pubDate><description><![CDATA[参考资料：


http://blog.youkuaiyun.com/shenxiaolu1984/article/details/51094959http://blog.youkuaiyun.com/zimenglan_sysu/article/details/52077138
http://blog.youkuaiyun.com/layumi1993/article/details/51854722
http://hy]]></description><category></category></item></channel></rss>