_menglan_Zi-CSDN博客

原创 SSD系列

1SSD: Single Shot MultiBox Detector. ECCV16 (arxiv)2R-SSD: enhancement of ssd by concatenating feature maps for object detection. BMVC17. (arxiv)3DSSD: Deconvolut

2017-12-28 10:05:43 2025 3

原创 human pose regression by combining indirect part detection and contextual information - arxiv - 1710

好久没写了，心血来潮好吧，怎么简单怎么来：human pose regression by combining indirect part detection and contextual information - arxiv - 1710.02322论文链接该论文的亮点在于重新用回了regression，而不是直接基于heat map的part detection方法。

2017-10-10 15:44:28 2421 1

原创 Improving Facial Attribute Prediction using Semantic Segmentation, in arxiv 2017.04

笔者最近在关注fine-grained方面的paper，发现有以下的方面去做：1 part-based2 weakly-supervised的，如second-orderless pooling（Compact Bilinear Pooling）等3 还是weakly-supervised的，但用上了proposals/grids/regions（如selective search）等

2017-05-08 22:34:18 1829

原创 Residual Attention Network for Image Classification, cvpr17

人至懒则无敌。cvpr17的论文，很有意思，值得一读和复现。（笔者懒，还是坐等开源吧）Residual Attention Network for Image Classification. In CVPR 2017.该模型的结构参考了: Stacked hourglass networks for human pose estimation. In arXiv 2016. （笔者

2017-05-08 22:00:04 7558 1

原创 Convolutional Low-Resolution Fine-Grained Classification. In arXiv, 2017.03

Convolutional Low-Resolution Fine-Grained Classification. In arXiv, 2017.03论文链接：Convolutional Low-Resolution Fine-Grained Classification.论文的idea很简单，就是把super resolution和classifier嵌入到一个模型中，并en

2017-03-20 11:03:30 1441

原创 Pyramid Scene Parsing Network - arxiv - 1612.01105

Pyramid Scene Parsing Network - arxiv - 1612.011052016年imagenet的scene parsing的冠军github: https://github.com/hszhao/PSPNetproject: http://appsrv.cse.cuhk.edu.hk/~hszhao/projects/pspnet/index.h

2016-12-09 11:06:43 3090 1

原创 Fully Convolutional Instance-aware Semantic Segmentation - arXiv 16.11

Fully Convolutional Instance-aware Semantic Segmentation - arXiv 16.11论文链接：https://arxiv.org/pdf/1611.07709v1.pdfgithub：https://github.com/daijifeng001/TA-FCN=====参考博文：1Instance-sensitive Fu

2016-11-29 12:30:04 3673

原创 Aggregated Residual Transformations for Deep Neural Networks - arxiv 16.11

Aggregated Residual Transformations for Deep Neural Networks, arXiv 16.11.论文链接: https://arxiv.org/abs/1611.05431=====这个是全明星阵容的paper，提出了深度模型另一个重要的需要考量的纬度：cardinality。(用简单的方式做不简单的事情）众所周知，提高模

2016-11-20 12:33:27 6549 1

原创 cv方面的博文收藏 -- 持续更新

（持续更新）- 深度学习变革视觉实例搜索 http://www.leiphone.com/news/201608/Iy4k4tU8zXThBYEQ.html

2016-11-15 15:37:14 1993

原创 speed improvement by merging batch normalization and scale

最近在看“在测试的时候, 怎么merge BN 和 Scale layer到conv layer”来加速。具体可以看：speed improvement by merging batch normalization and scale #5

2016-11-04 10:19:39 1279 2

原创 Deep Image Retrieval: Learning global representations for image search. In ECCV, 2016.

Deep Image Retrieval: Learning global representations for image search. In ECCV, 2016.论文地址：https://arxiv.org/abs/1604.01325extended version：end to end learning of deep visual representations for i

2016-11-03 17:15:47 5699 4

原创 A Discriminative Feature Learning Approach for Deep Face Recognition, ECCV16.

A Discriminative Feature Learning Approach for Deep Face Recognition, ECCV16.论文：http://ydwen.github.io/papers/WenECCV16.pdf代码（caffe）：https://github.com/kpzhang93/caffe-face代码（mxnet）：https://gith

2016-10-29 10:45:01 3212

原创 StuffNet: Using ‘Stuff’ to Improve Object Detection - arxiv 16.10

StuffNet: Using ‘Stuff’ to Improve Object Detection, Arxiv 16.10论文地址：https://arxiv.org/pdf/1610.05861v1.pdf类似的论文：Contextual Priming and Feedback forFaster R-CNN, eccv16.=====好像现在都喜欢用multi-task

2016-10-27 20:10:40 1650

原创 plateau lr policy

好久没有冒泡了，哈哈。plateau lr policy来自于PVANET: Deep but Lightweight Neural Networks for Real-time Object Detection这篇论文。这里从issue里面整理出来的，欢迎大家指点，以及如果你用了plateau lr policy来做实验，是否可以告知下效果如何？issues:Add "plate

2016-10-13 10:05:16 2784 2

原创 PVANET: Deep but Lightweight Neural Networks for Real-time Object Detection - arxiv 2016.08

PVANET: Deep but Lightweight Neural Networks forReal-time Object Detection, arxiv 16.08论文地址：https://arxiv.org/pdf/1608.08021v1.pdf先看leaderboard再看下speed厉害的不要不要的。=====转正题，下面介绍下这篇

2016-09-08 15:07:33 6489 12

原创 Instance-sensitive Fully Convolutional Networks - eccv 2016

Instance-sensitive Fully Convolutional Networks - eccv 2016=====论文地址：http://arxiv.org/abs/1603.08678=====一句话概括：根据local coherence的特性，以sliding window的方式，利用FCN产生positive-sensitive的instance-le

2016-09-06 17:36:16 5243 3

原创 R-FCN: Object Detection via Region-based Fully Convolutional Networks (NIPS 16), Arxiv 16.05

R-FCN: Object Detection via Region-based Fully Convolutional Networks, Arxiv 16.05=====blogs：论文阅读]R-FCN: Object Detection via Region-based Fully Convolutional Networks： http://blog

2016-08-31 13:59:47 5027 4

原创 - UnitBox An Advanced Object Detection Network，arxiv 16.08

- UnitBox An Advanced Object Detection Network，arxiv 16.08 （download）该论文提出了一种新的loss function：IoU loss。这点比较有意思，也容易复现。 ====== 论文分析了faster-rcnn和densebox的优缺点： 1 fast

2016-08-16 21:43:33 5683 1

原创 Bootstrapping Face Detection with Hard Negative Examples，arxiv 16.08

- Bootstrapping Face Detection with Hard Negative Examples，arxiv 16.08（download）昨天在微博上看到这个论文的时候，很是惊讶，为啥？因为出自小米的为发烧而生（跑分） ===== 论文比较短，主要提出了一个hard negative mining的方式，以及基于faste

2016-08-15 19:43:29 3366 4

原创 Multi-view Face Detection Using Deep Convolutional Neural Networks, ICMR 15.

Multi-view Face Detection Using Deep Convolutional Neural Networks, ICMR 15. 将Alexnet转成FCN的形式也就是227*227的输入会输出1*1的heat map），将detection问题转为classification问题。很简单，对每个region/propos

2016-08-11 20:21:37 3051 8

原创 convolutional pose machines, CVPR 2016

convolutional pose machines, CVPR 2016.论文：http://arxiv.org/abs/1602.00134project：https://github.com/shihenw/convolutional-pose-machines-release=====看到这篇论文，心里还是有点难受的，当初做pose的毕设时，为什么就不参考下这篇掉渣天的论

2016-07-31 10:21:13 10594 7

原创 DeepFashion: Powering Robust Clothes Recognition and Retrieval with Rich Annotations – CVPR 2016

DeepFashion: Powering Robust ClothesRecognition and Retrieval with Rich Annotations – CVPR 2016论文&project：http://personal.ie.cuhk.edu.hk/~lz013/projects/DeepFashion.html1 clothes的chall

2016-07-03 21:49:17 6149 13

原创 1080P 显卡安装过程

1080P 显卡安装过程版权归 Deep Learning Team, Malong Technologies 所有！www.malongtech.cn (p.s. we’re hiring!) 安装过程（successfully）1 安装ubuntu14.04。（可选择14.04或者16.0

2016-07-02 13:28:24 4589 11

原创 Network in Network - ICLR 2014

NIN是2014 ICLR的一篇论文，你会发现很多论文都会引用到该论文，可见该论文是牛逼的。之所以想看这篇，是因为自己想弄到object classification & detection的经典的network architecture的脉络（说白了，笔者菜鸟一个，哭去，连这篇论文都没看过），这里可以推荐各位看官看下这篇 Neural Network Architectures（h

2016-06-13 18:54:10 3530

转载论文笔记《Learning Multi-Domain Convolutional Neural Networks for Visual Tracking》

转载于：http://backnode.github.io/pages/2015/11/02/CNN-for-tracking.html版权归原作者所有论文笔记《Learning Multi-Domain Convolutional Neural Networks for Visual Tracking》0. 摘要基于CNN的一个跟踪

2016-06-06 11:10:15 2484

原创 Let there be Color!: Automatic Image Colorization with Simultaneous Classification - sig - 2016

最近在arxiv上逛逛，看到一篇关于colorization的paper觉得挺有意思的：Let there be Color!: Joint End-to-end Learning of Global and Local Image Priors for Automatic Image Colorization with Simultaneous Classification - sig -

2016-05-05 10:55:24 3431 1

原创 Training Region-based Object Detectors with Online Hard Example Mining - cvpr 2016 oral

这是rgb大神的又一神作，cvpr 2016的oral paper，论文地址here。先上图，如Figure2所示：从图中可以看出，本文的亮点在于在每次迭代中，较少训练样本下，如何hard negative mining，来提升效果。即针对Fast-RCNN框架，在每次minibatch（1张或者2张）训练时加入在线筛选hard region的策略，达到新的SoA。

2016-05-04 21:37:46 10291 38

转载【转载】论文笔记《Object detection via a multi-region & semantic segmentation-aware CNN model》

转载于：http://zhangliliang.com/2015/05/17/paper-note-ms-cnn/版权归原作者所有********************************************************************论文笔记《Object detection via a multi-region & semanti

2016-05-04 18:51:44 4929 1

原创 Pose-Action-Caffe

等我弄完毕设，就来完善、整理自己基于fast-rcnn来弄的pose estimation 和 action recognition的方法。现在已经把代码放到github上了，欢迎各位看客前来指教。 pose-action-caffepose的参考论文： 1 Flowing ConvNets for Human Pose Estimation in Videos, ICCV

2016-04-07 11:54:29 2318 4

原创 research of action recognition and classfication in still images

一篇关于静态图像的活动/动作识别的调研文档，具体看以下连接https://github.com/zimenglan-sysu-512/paper-note/blob/master/action-recognition/research.of.action.classfication.or.recognition.in.static.image.pdf请各位看官不要嫌弃。

2016-03-03 19:31:20 903

原创求两帧图像或着视频的光流（optical flow）

最近在弄视频方面的pose estimation，由于不才，没有想到很高大上的idea来利用视频的信息（如LSTM等），只想到用optical flow的方式来利用视频信息，（路过的不要鄙视鄙人）至于怎么弄pose，欢迎各位看客前来赐教。至于怎么产生optical flow，请各位看官，移步至下面的连接https://github.com/zimenglan-sysu-51

2016-03-02 21:39:58 11994 3

转载论文提要“You Only Look Once: Unified, Real-Time Object Detection”

转载于:http://blog.csdn.net/cv_family_z/article/details/46803421版权归原作者所有.项目主页：http://pjreddie.com/darknet/yolo/这篇文章着重在检测的速度提升，区别于之前的方法是用分类器来做检测，文章对bbox和对应的类概率进行回归，检测速度可以达到45f/s，mAP与原始的R-CNN差不多。

2016-01-10 15:44:47 1601

原创 Heterogeneous Multi-task Learning for Human Pose Estimation with Deep Convolutional Neural Network

Heterogeneous Multi-task Learning for Human Pose Estimation with Deep Convolutional Neural Network论文题目Heterogeneous Multi-task Learning for Human Pose Estimation with Deep Convolutional Neural

2015-12-14 14:28:09 2280 2

原创 Finding Action Tubes - cvpr - 2015

论文题目Finding Action Tubes, 论文链接该篇论文是CVPR 2015的, 主要讲述了action tube的localization.直接看图说话, 该论文的核心思想/步骤可以分为两个components: 1 Action detection at every frame of the video 2 Linked detection

2015-11-25 00:19:10 1838

H-Index原题Given an array of citations (each citation is a non-negative integer) of a researcher, write a function to compute the researcher's h-index.According to the definition of h-index on Wikipedia:

2015-11-14 09:47:54 1683

原创 Modeling Video Evolution For Action Recognition - cvpr - 2015

Modeling Video Evolution For Action Recognition - cvpr - 2015论文题目Modeling Video Evolution For Action Recognition, 链接该篇论文是CVPR 2015的, 文章的核心是feature descriptors的获取. 直接看图说话:

2015-11-13 18:59:57 1919 4

原创 P-CNN: Pose-based CNN Features for Action Recognition (CNN篇)

论文题目P-CNN: Pose-based CNN Features for Action Recognition, 链接之前看过在静态图像上做action recognition的论文, 如Georgia Gkioxari这位大美女的论文, Contextual Action Recognition with R*CNN (可以看柱锦师兄的blog)和RGB视频里面的

2015-11-12 18:06:19 5748 11

原创 Range Sum Query - Immutable

题目链接原题Given an integer array nums, find the sum of theelements between indices i and j (i ≤ j), inclusive.Example:Given nums = [-2, 0, 3, -5, 2, -1]sumRange(0, 2) -> 1sumRange(2, 5) -> -1sumRange(0

2015-11-12 12:03:43 611

原创 Range Sum Query 2D - Immutable

题目链接原题Given a 2D matrix matrix, find the sum of the elements inside the rectangle defined by (row1, col1), (row2, col2).Example:Given matrix = [ [3, 0, 1, 4, 2], [5, 6, 3, 2, 1], [1, 2, 0, 1, 5

2015-11-12 12:00:11 746

原创 Different Ways to Add Parentheses

题目链接原题Given a string of numbers and operators, return all possible results from computing all the different possible ways to group numbers and operators. The valid operators are +, - and *.思路左右子串分别计算

2015-11-05 23:52:24 534

空空如也

空空如也