<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[CUBE的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/qq_26074263</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; qq_26074263]]></copyright><item><title><![CDATA[【Projects】Facebook开源Blender! 浅谈ParlAI中的Chit-chat任务【一】]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/105954326</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/105954326</guid><author>qq_26074263</author><pubDate>Wed, 06 May 2020 20:57:37 +0800</pubDate><description><![CDATA[此项目链接为ParlAI Projects主页，列出了基于ParlAI框架实现的各类projects。本篇文章则主要介绍Chit-chat任务中，基于ParlAI的模型表现，以及使用方法。
QuickStart
Requirements

python 3.6
torchtext

Steps

clone

git clone https://github.com/facebookresearc...]]></description><category></category></item><item><title><![CDATA[【PaperNotes】视频分类【二】监督式端到端CNN系列]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/105897573</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/105897573</guid><author>qq_26074263</author><pubDate>Wed, 06 May 2020 11:32:24 +0800</pubDate><description><![CDATA[（1） 监督学习方法
i. 基于图像的视频分类：将视频片段视为视频帧的集合，每个视频帧的特征通过ImageNet数据集上预先训练的最高水平的深度模型（如AlexNet，VGGNet，GoogLeNet，ResNet）进行获取。最终，帧层特征汇聚为视频层特征，作为标准分类器（如SVM）识别的输入。
ii. 端到端的CNN网络：关注于利用CNN模型学习视频隐含的时空模式，如3D CNN，Two-str...]]></description><category></category></item><item><title><![CDATA[【PaperNotes】视频分类【一】]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/105884591</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/105884591</guid><author>qq_26074263</author><pubDate>Sat, 02 May 2020 01:32:06 +0800</pubDate><description><![CDATA[Appearance-and-Relation Networks for Video Classification
简介

提出ARTNet，学习端到端的视频表征
ARTNets是由多个SMART blocks堆叠而来的，SMART可用于同时从RGB输入中建模出appearance与relation
SMART将spatiotemporal学习分为两部分：用于空间建模的appearance分支与用...]]></description><category></category></item><item><title><![CDATA[【PaperNotes】时间交错网络Temporal Interlacing Network]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/105861440</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/105861440</guid><author>qq_26074263</author><pubDate>Thu, 30 Apr 2020 22:52:43 +0800</pubDate><description><![CDATA[论文链接
相关知识点
Temporal Models
时间模型依赖于时间相关的状态与sensor变量如下：
Xt:时间t下的不可直接观测状态集合
Et:时间t下的可观测的迹象变量
Markov假设： Xt依赖于X0:t-1的有限子集
First-order Markov过程： P(Xt|X0:t-1) = P(Xt|Xt-1)
Second-order Markov过程：P(Xt|X0:t-1) ...]]></description><category></category></item><item><title><![CDATA[SimpleDet安装使用流程教学]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/87611349</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/87611349</guid><author>qq_26074263</author><pubDate>Mon, 18 Feb 2019 12:28:40 +0800</pubDate><description><![CDATA[SimpleDet安装使用流程教学SimpleDet简介mxnext安装
SimpleDet简介

A Simple and Versatile Framework for Object Detection and Instance Recognition

simpledet github仓库地址
mxnext安装

setup mxnext, a wrapper of mxnet symbol...]]></description><category></category></item><item><title><![CDATA[SNIPER 获取固定参数并创建模块]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/86648512</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/86648512</guid><author>qq_26074263</author><pubDate>Fri, 25 Jan 2019 16:40:08 +0800</pubDate><description><![CDATA[Get list of fixed parameters
    sym_inst = eval('{}.{}'.format(config.symbol, config.symbol))(n_proposals=400, momentum=args.momentum)
    sym = sym_inst.get_symbol_rcnn(config)
    fixed_param_names...]]></description><category></category></item><item><title><![CDATA[SNIPER MNIterator代码解析]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/86647409</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/86647409</guid><author>qq_26074263</author><pubDate>Fri, 25 Jan 2019 15:51:32 +0800</pubDate><description><![CDATA[main_train.py
    train_iter = MNIteratorE2E(roidb=roidb, config=config, batch_size=batch_size, nGPUs=nGPUs, threads=config.TRAIN.NUM_THREAD, pad_rois_to=400)

SNIPER/lib/iterators
MNIteratorE2E
class...]]></description><category></category></item><item><title><![CDATA[SNIPER/lib/data_utils load_data加载数据]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/86630696</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/86630696</guid><author>qq_26074263</author><pubDate>Thu, 24 Jan 2019 17:37:58 +0800</pubDate><description><![CDATA[

def load_gt_roidb(dataset_name, image_set_name, root_path, dataset_path, result_path=None, flip=False):


def load_proposal_roidb(dataset_name, image_set_name, root_path, dataset_path, result_path=N...]]></description><category></category></item><item><title><![CDATA[Efficient Multi-Scale Training训练代码解析]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/86627836</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/86627836</guid><author>qq_26074263</author><pubDate>Thu, 24 Jan 2019 15:34:00 +0800</pubDate><description><![CDATA[Efficient Multi-Scale Training训练代码解析

def parser() 使用argparse模块实现命令行解析[1]


导入argparse模块
创建解析器对象ArgumentParser，可添加参数
add_argument()，指定程序需要接受的命令参数

add_argument()定位参数（必选）:
parser.add_argument(“echo”, h...]]></description><category></category></item><item><title><![CDATA[转换为COCO数据格式代码]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/86626833</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/86626833</guid><author>qq_26074263</author><pubDate>Thu, 24 Jan 2019 14:53:14 +0800</pubDate><description><![CDATA[COCO数据格式是许多框架与网络结构通用的数据格式，这里给出一般标注的图片数据格式转换为COCO的数据格式，存储为json文件。
存储主干部分
dataset = { 'info': {
            'description': &amp;amp;amp;quot;This is stable 1.0 version of the 2014 MS COCO dataset.&amp;amp;amp;quot;, 
            'url': ...]]></description><category></category></item><item><title><![CDATA[CHAPTER1 Geometric Camera Models未完成]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/86617863</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/86617863</guid><author>qq_26074263</author><pubDate>Wed, 23 Jan 2019 21:22:00 +0800</pubDate><description><![CDATA[Geometric Camera Models
1.1 Image Formation
1.1.1 Pinhole Perspective
1.1.2 Weak Perspective
1.1.3 Cameras with Lenses
1.1.4 The Human Eye
1.2 Intrinsic and Extrinsic Parameters
1.2.1 Rigid Transforma...]]></description><category></category></item><item><title><![CDATA[计算机视觉初学者必知名词与概念Terms & Concepts for beginners in Computer Vision]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/86175620</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/86175620</guid><author>qq_26074263</author><pubDate>Wed, 09 Jan 2019 20:07:59 +0800</pubDate><description><![CDATA[25个深度学习初学者必知词汇与概念
作为一个CV领域的初学者，看论文时常对一些术语不熟悉或者理解不深刻，这里记录一些阅读论文过程中的terms &amp;amp;amp;amp;amp; concepts，加深记忆
weight-sharing 共享权重 : use fixed weights for the filter across the entire input
反向传播中的weight-sharing
updating...]]></description><category></category></item><item><title><![CDATA[复杂验证码识别APP]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/86076072</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/86076072</guid><author>qq_26074263</author><pubDate>Tue, 08 Jan 2019 14:52:56 +0800</pubDate><description><![CDATA[项目地址 https://github.com/cubeFUN/AppSRTP
updating…

]]></description><category></category></item><item><title><![CDATA[基于MVVM模式的复杂验证码识别]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/86075969</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/86075969</guid><author>qq_26074263</author><pubDate>Tue, 08 Jan 2019 14:51:37 +0800</pubDate><description><![CDATA[项目地址 https://github.com/cubeFUN/CAPTCHA_MVVM
updating…

]]></description><category></category></item><item><title><![CDATA[SNIPER实践于自动驾驶企业项目-实时障碍物检测]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/86075713</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/86075713</guid><author>qq_26074263</author><pubDate>Tue, 08 Jan 2019 14:48:02 +0800</pubDate><description><![CDATA[原论文 https://arxiv.org/abs/1805.09300
项目地址 https://github.com/cubeFUN/Driving-Automation
update…

]]></description><category></category></item><item><title><![CDATA[Apollo无人车驾驶技术之感知]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/82795704</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/82795704</guid><author>qq_26074263</author><pubDate>Fri, 21 Sep 2018 01:40:50 +0800</pubDate><description><![CDATA[感知部分
Apollo 将ROI过滤器应用于点云和图像数据 -&amp;gt; 缩小范围加快感知

通过监测网络馈送以过滤的点云
输出用语构建围绕对象的三位框架
Point Cloud -&amp;gt; Detection Network -&amp;gt; Center Offset; Objectness; Positiveness; Object Height; Class Probability
Which l...]]></description><category></category></item><item><title><![CDATA[Udacity&Baidu无人驾驶概览]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/82779872</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/82779872</guid><author>qq_26074263</author><pubDate>Wed, 19 Sep 2018 23:12:41 +0800</pubDate><description><![CDATA[Levels of Driving Automation

Level1 Driver Assistance Driver Fully Engaged
Level2 Partial Automation Automatic Cruise Control Automatic Lane Keeping
Level3 Conditional Automation Human Take Over When...]]></description><category></category></item><item><title><![CDATA[媒体信号处理Digital Multimedia Note1]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/82758970</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/82758970</guid><author>qq_26074263</author><pubDate>Tue, 18 Sep 2018 15:48:08 +0800</pubDate><description><![CDATA[Digital Multimedia Note1内容提要信号的模数转换信息论基本知识
内容提要
信号的模数转换
信息论基本知识
PCM与DPCM编码
。。。
信号的模数转换
信号采样频率
信号量化
信噪比，信号量化噪声比
麦克风：典型模数转换设备
模拟信号是在时间和空间上的电平信号

关键问题

采样频率
每一个采样点得到的信号值如何转换为整数值（量化）往上or往下取整
如何评价模拟-》数字信号转...]]></description><category></category></item><item><title><![CDATA[如何编写 C++ 游戏引擎]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/81284252</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/81284252</guid><author>qq_26074263</author><pubDate>Mon, 30 Jul 2018 14:17:20 +0800</pubDate><description><![CDATA[如何编写 C++ 游戏引擎

原创： 伯乐在线  CPP开发者  6月5日 
（点击上方公众号，可快速关注）

编译：伯乐在线 - 李大萌  英文：Jeff Preshing 
http://blog.jobbole.com/113960/

最近我在用 C++ 写游戏引擎，再用这个引擎做了一个移动端小游戏跳一跳（Hop Out）。下面是截自我的 iPhone6 的一个小片段。

跳一跳是我想玩的...]]></description><category></category></item><item><title><![CDATA[CMU最新视觉特征自监督学习模型——TextTopicNet]]></title><link>https://blog.youkuaiyun.com/qq_26074263/article/details/81277630</link><guid>https://blog.youkuaiyun.com/qq_26074263/article/details/81277630</guid><author>qq_26074263</author><pubDate>Mon, 30 Jul 2018 00:08:50 +0800</pubDate><description><![CDATA[CMU最新视觉特征自监督学习模型——TextTopicNet

原创： 专注技术分享的  人工智能头条  今天

作者 | Yash Patel，Lluis Gomez，Raul Gomez，Marcal Rusinol，Dimosthenis Karatzas， C.V. Jawahar 
译者 | 林椿眄 
编辑 | Jane 
出品 | 人工智能头条

▌摘要

深度学习方法在计算机视觉领域...]]></description><category></category></item></channel></rss>