<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[nanxiaoting的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/nanxiaoting</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; nanxiaoting]]></copyright><item><title><![CDATA[求最大字串和 （在线更新法）]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/85269941</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/85269941</guid><author>nanxiaoting</author><pubDate>Wed, 26 Dec 2018 22:32:36 +0800</pubDate><description><![CDATA[

public class MaxSubArray {
	public static void main(String[] args) {
		int[] arr = {-2,1,-3,4,-1,2,1,-5,4};
		int[] arr1={-1,2,3,50,-16,1,20,-12};
		maxarr(arr1);
	}
	public static void maxarr(int[]...]]></description><category></category></item><item><title><![CDATA[coco 关键点json文件解析]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/85245860</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/85245860</guid><author>nanxiaoting</author><pubDate>Tue, 25 Dec 2018 12:04:19 +0800</pubDate><description><![CDATA[COCO是微软提供的一个图像识别的数据集。其中包括3个tasks，分别是object instances, object keypoints, 和image captions，存储格式为JSON。

基本的数据结构定义如下：


{
&quot;info&quot; : info, &quot;images&quot; : [image], &quot;annotations&quot; : [annotation], &quot;licenses&quot; : [lice...]]></description><category></category></item><item><title><![CDATA[python  pandas（一）]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/84072967</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/84072967</guid><author>nanxiaoting</author><pubDate>Wed, 14 Nov 2018 20:02:58 +0800</pubDate><description><![CDATA[
#coding=utf-8
import numpy as np
import pandas as pd

df1 = pd.DataFrame(pd.read_csv('./house_data/all/test.csv',header=1))   # 读取.csv文件
print df1.shape

df2 = pd.DataFrame({&quot;id&quot;:[1002,1001,1003,1004...]]></description><category></category></item><item><title><![CDATA[pytorch 实践二]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/83903450</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/83903450</guid><author>nanxiaoting</author><pubDate>Fri, 09 Nov 2018 21:17:21 +0800</pubDate><description><![CDATA[
#coding=utf-8
import torch
import numpy as np
import torch.nn as nn
from torchvision import datasets, transforms
from torch.utils import data
import os
from PIL import Image
import torch.nn.functiona...]]></description><category></category></item><item><title><![CDATA[pytorch自我学习基础（一） 线性回归]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/83855682</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/83855682</guid><author>nanxiaoting</author><pubDate>Thu, 08 Nov 2018 09:20:14 +0800</pubDate><description><![CDATA[pytorch 版本 0.2.0


#coding=utf-8
from matplotlib import pylab as plt
import numpy as np
import random
import torch.nn as nn
import torch
from torch.autograd import Variable

num_inputs = 2
num_example...]]></description><category></category></item><item><title><![CDATA[faster rcnn理解]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/83143773</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/83143773</guid><author>nanxiaoting</author><pubDate>Thu, 18 Oct 2018 11:24:06 +0800</pubDate><description><![CDATA[
	AnchorTargetCreator ： 负责在训练 RPN 的时候，从上万个 anchor 中选择一些 (比如 256) 进行训练，以使得正负样本比例大概是 1:1. 同时给出训练的位置参数目标。 即返回gt_rpn_loc和gt_rpn_label。
	
	
	ProposalTargetCreator： 负责在训练 RoIHead/Fast R-CNN 的时候，从 RoIs 选择一部分...]]></description><category></category></item><item><title><![CDATA[训练faster-rcnn python版本的时候报错No module named cython_bbox]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/83107766</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/83107766</guid><author>nanxiaoting</author><pubDate>Wed, 17 Oct 2018 15:06:02 +0800</pubDate><description><![CDATA[

补充一点：将proposal_target_layer.py中的


from utils.cython_bbox import bbox_overlaps

替换成：


from utils.bbox import bbox_overlaps ]]></description><category></category></item><item><title><![CDATA[terminal中可以import caffe，但是pycharm中无法import  及faster rcnn包标红]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/83014941</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/83014941</guid><author>nanxiaoting</author><pubDate>Thu, 11 Oct 2018 16:46:01 +0800</pubDate><description><![CDATA[亲测有效

http://askubuntu.com/questions/684550/importing-a-python-module-works-from-command-line-but-not-from-pycharm

按照图片说明进行，在最后一步 添加路径时候选择到自己的如下路径中

 


 &quot;/your/caffe/path/caffe-master/python&quot;


同理，对...]]></description><category></category></item><item><title><![CDATA[批量删除VOC的XML中的某些节点]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/82859422</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/82859422</guid><author>nanxiaoting</author><pubDate>Wed, 26 Sep 2018 21:10:51 +0800</pubDate><description><![CDATA[
import xml.etree.cElementTree as ET
import os
path_root = ['E:\data-VOC0712\VOC2007\Annotations',
             'E:\data-VOC0712\VOC2012\Annotations']

CLASSES = [
           &quot;bottle&quot;,
           &quot;cat...]]></description><category></category></item><item><title><![CDATA[python 数据库 mongodb]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/82721083</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/82721083</guid><author>nanxiaoting</author><pubDate>Sun, 16 Sep 2018 10:16:41 +0800</pubDate><description><![CDATA[
#coding=utf-8

import pymongo
#print pymongo.__version__

client = pymongo.MongoClient(host='127.0.0.1',port=27017)
db = client.test
collection = db.students
student = {
    'id':'20170905',
    'nam...]]></description><category></category></item><item><title><![CDATA[mysql python]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/82718224</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/82718224</guid><author>nanxiaoting</author><pubDate>Sat, 15 Sep 2018 21:51:18 +0800</pubDate><description><![CDATA[
#coding=utf-8

import MySQLdb

con = MySQLdb.connect(host='127.0.0.1',user='root',passwd='2',db='nan',port=3306)
cur = con.cursor()
cur.execute('CREATE TABLE IF NOT EXISTS tableone (id int not null a...]]></description><category></category></item><item><title><![CDATA[数据库   （一） sqlite3]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/82708851</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/82708851</guid><author>nanxiaoting</author><pubDate>Sat, 15 Sep 2018 09:45:51 +0800</pubDate><description><![CDATA[
#coding=utf-8
import sqlite3
import sys
reload(sys)
sys.setdefaultencoding('utf-8')

con = sqlite3.connect('/home/flyvideo/sqlite3database.db')
con.text_factory = str # 插入的字符串中含有中文字符,得加这一行
con.execut...]]></description><category></category></item><item><title><![CDATA[文字识别——检测部分  CTPN论文翻译]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/82695963</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/82695963</guid><author>nanxiaoting</author><pubDate>Thu, 13 Sep 2018 22:24:02 +0800</pubDate><description><![CDATA[原文翻译地址：https://blog.youkuaiyun.com/quincuntial/article/details/79475339

翻译论文汇总：https://github.com/SnailTyan/deep-learning-papers-translation

Detecting Text in Natural Image with Connectionist Text Proposa...]]></description><category></category></item><item><title><![CDATA[文字识别开源库]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/82562877</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/82562877</guid><author>nanxiaoting</author><pubDate>Sun, 09 Sep 2018 21:47:44 +0800</pubDate><description><![CDATA[一、SWT识别：
yestinsong/Text-Detection( Text Detection System with MSER , SWT and Text Verification(fft and pca) )
https://github.com/yestinsong/Text-Detection
aperrau/DetectText：Detect text with stroke w...]]></description><category></category></item><item><title><![CDATA[白翔2018Mask TextSpotter: An End-to-End Trainable Neural Network for Spotting Text with Arbitrary Shap]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/82530153</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/82530153</guid><author>nanxiaoting</author><pubDate>Sat, 08 Sep 2018 15:52:01 +0800</pubDate><description><![CDATA[转载的别人的，主要是方便自己看

引入Mask R-CNN思想通过语义分割进行任意形状文本检测与识别。

 

华中科技大学白翔老师团队在自然场景文本检测与识别领域成果颇丰，这篇被ECCV2018接收的论文《Mask TextSpotter: An End-to-End Trainable Neural Network for Spotting Text with Arbitrary Shapes...]]></description><category></category></item><item><title><![CDATA[文字识别论文合集（主要转自其他的博客，记录一下）]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/82528257</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/82528257</guid><author>nanxiaoting</author><pubDate>Sat, 08 Sep 2018 12:06:36 +0800</pubDate><description><![CDATA[原文链接：https://blog.youkuaiyun.com/peaceinmind/article/details/51387367

本文写成时主要参考了[1,2], 后面加了一些自己收集的，不过大家都在更新，所以区别点击打开链接不是很大~

 

综述

[2015-PAMI-Overview]Text Detection and Recognition in Imagery: A Survey[p...]]></description><category></category></item><item><title><![CDATA[Yolo v3 详解]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/82497731</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/82497731</guid><author>nanxiaoting</author><pubDate>Fri, 07 Sep 2018 15:43:14 +0800</pubDate><description><![CDATA[原文链接：https://blog.youkuaiyun.com/u014380165/article/details/80202337#commentBox

论文：YOLOv3: An Incremental Improvement 
论文地址：https://pjreddie.com/media/files/papers/YOLOv3.pdf

YOLO系列的目标检测算法真的非常赞！这篇博客就来介绍YO...]]></description><category></category></item><item><title><![CDATA[Yolo V2理解]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/82469823</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/82469823</guid><author>nanxiaoting</author><pubDate>Thu, 06 Sep 2018 23:17:39 +0800</pubDate><description><![CDATA[原文：https://blog.youkuaiyun.com/u014380165/article/details/77961414

论文： YOLO9000：Better，Faster，Stronger 
论文链接：https://arxiv.org/abs/1612.08242

YOLO9000是CVPR2017的最佳论文提名。首先讲一下这篇文章一共介绍了YOLO v2和YOLO9000两个模型，二者...]]></description><category></category></item><item><title><![CDATA[遇到usr/bin/ld: cannot find -lcufft解决方法。]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/82468272</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/82468272</guid><author>nanxiaoting</author><pubDate>Thu, 06 Sep 2018 21:06:03 +0800</pubDate><description><![CDATA[遇到usr/bin/ld: cannot find -lcufft解决方法。

原文：https://blog.youkuaiyun.com/wan_exe/article/details/70171052

版权声明：本文为博主原创文章，未经博主允许不得转载。 https://blog.youkuaiyun.com/WAN_EXE/article/details/70171052

Ubuntu15.04中安装完ope...]]></description><category></category></item><item><title><![CDATA[RoI Pooling详解]]></title><link>https://blog.youkuaiyun.com/nanxiaoting/article/details/82421076</link><guid>https://blog.youkuaiyun.com/nanxiaoting/article/details/82421076</guid><author>nanxiaoting</author><pubDate>Wed, 05 Sep 2018 11:50:43 +0800</pubDate><description><![CDATA[ RoI Pooling详解

RoI Pooling其实是SPP pooling的一种特殊情况。SPP pooling可以采用多个金字塔层，RoI只采用一种。具体做法为: 对于映射到feature map上的RoI(Region Proposal通过变换得来)，假设该RoI大小为h×wh×w,RoI Pooling的输出为H×WH×W(H,WH,W为超参数)。将h×wh×w的RoIRoI分解成H...]]></description><category></category></item></channel></rss>