gaussianface源码

① 先驱者(5) AI巨头如何跨界自动驾驶

[汽车之家技术]有这样一家成立于2014年，旋即成为该领域“独角兽”的高科技企业，说起它的名字大部分人可能以为是历史课本某一章节的名称，但不夸张的说，如今我们几乎每天都离不开它的产品或技术。它赋予机器超过人类的辨别能力，并且在还会创造出更多来自于科幻小说的未来场景，那么TA与我们的《先驱者》系列内容又有什么联系呢？且听我娓娓道来。

写在最后：

往期回顾：

先驱者（4）激光大战中的中国高端玩家

先驱者（3）中国激光雷达企业逆袭记

先驱者（2）机器视觉领域的中国挑战者

先驱者（1）一文了解世界自动驾驶格局

② 彩色图像分割源码

能不能给我发一份呀？

③ 在like face上面看见了高斯的防晒喷雾好想好好评的，不过不知道是不是正品呢~

oq
468

④ linux gaussian怎么安装linda

一、 背景介绍
Gaussian是目前计算化学领域内最流行、应用范围最广的商业化量子化学计算程序包。它最早是由美国卡内基梅隆大学的约翰·波普（John A Pople, 1998年诺贝尔化学奖）在60年度末、70年代初主导开发的。其名称来自于该软件中所使用的高斯型基组。最初，Gaussian的着作权属于约翰·波普供职的卡内基梅隆大学；1986年，约翰·波普进入美国西北大学后，其版权由Gaussian,Inc.公司所持有。Gaussian软件的出现降低了量子化学计算的门槛，使得从头计算方法可以广泛使用，从而极大地推动了其在方法学上的进展。
到目前为止，Gaussian已经推出了12个版本，包括Gaussian70、Gaussian76、Gaussian80、Gaussian82、Gaussian86、Gaussian88、Gaussian90、Gaussian92、Gaussian92/DFT、Gaussian94、Gaussian98、Gaussian03等，其版本数字也是该版本发布的年份。其中，每个版本发布后，还陆续发布了一些这些版本的修订版。目前最新的版本是Gaussian03 Revision D.01/D.02。
Gaussian程序是用FORTRAN语言编写的，它从量子力学的基本原理出发，可计算能量、分子结构、分子体系的振动频率以及大量从这些基本计算方法中导出的分子性质。它能用于研究不同条件下的分子和反应，包括稳定的粒子和实验上难以观测的化合物，例如瞬时的反应中间物和过渡结构。
Gaussian的并行模式是采用OpenMP来实现的。OPENMP的并行实现是针对共享内存的机器的，实现方法简单。因此Gaussian在共享内存的机器上，能获得很好的性能。对于跨节点的计算，Gaussian使用TCP Linda软件来实现。TCP Linda是一个虚拟共享内存的并行执行环境，它可以把一个通过网络连接的分布式内存的机群或工作站虚拟成共享内存环境，从而使像Gaussian这样的用OPENMP实现并行的程序能够在分布式内存的机器上运行。

二、 软件的安装设置
1、将压缩包解开
# tar zxf OPT-900N.taz //g03 E01以上版本支持上海处理器 （可以查看文件日期在2007年以后的）

2、准备环境变量文件g03.sh

放入g03源代码目录，如/home/users/mjhe/g03/

#cat g03.sh

g03root="/home/users/mjhe"

GAUSS_SCRDIR="/scratch"

export g03root GAUSS_SCRDIR

. $g03root/g03/bsd/g03.profile

3、准备运行脚本
放入算例目录，如g03test

> cat g03.pbs

###########################################################################

# Script for submitting parallel Gaussian 03 jobs to Dawning cluster.

# Lines that begin with #PBS are PBS directives (not comments).

# True comments begin with "# " (i,e., # followed by a space).

###########################################################################

#PBS -S /bin/bash

#PBS -N gaussian

#PBS -j oe

#PBS -l nodes=1:ppn=8

##PBS -l walltime=860:00:00

#PBS -V

##PBS -q middle

#############################################################################

# -S: shell the job will run under

# -o: name of the queue error filename

# -j: merges stdout and stderr to the same file

# -l: resources required by the job: number of nodes and processors per node

# -l: resources required by the job: maximun job time length

#############################################################################

INFILE=$file

# Define variable "jobname".

jobname=`echo $INFILE | awk -F. '{printf $1}'`

username=`whoami`

# Define the location where Gaussian was installed and run a setup script, g03.profile.

g03root="/data/users/ceszhcy/"

GAUSS_SCRDIR="/state/partition1/tmp/"

export g03root GAUSS_SCRDIR

. $g03root/g03/bsd/g03.profile

# Make a directory in scr and .com and .g03 file to there.

GAUSS_RUNDIR=${GAUSS_SCRDIR}/${username}.${PBS_JOBID}

if [ ! -a $GAUSS_RUNDIR ]; then

echo "Scratch directory $GAUSS_RUNDIR created."

mkdir -p $GAUSS_RUNDIR

fi

cp $PBS_O_WORKDIR/${jobname}.* $GAUSS_RUNDIR

ORIG_PBS_O_WORKDIR=${PBS_O_WORKDIR}

cd $PBS_O_WORKDIR

# Setup for Gaussian 03:

# =======================

# Make a scratch directory if it doesn't already exist.

GAUSS_SCRDIR=${GAUSS_SCRDIR}/${username}.${PBS_JOBID}/${jobname}

if [ ! -a $GAUSS_SCRDIR ]; then

echo "Scratch directory $GAUSS_SCRDIR created."

mkdir -p $GAUSS_SCRDIR

fi

export GAUSS_SCRDIR

echo "Using $GAUSS_SCRDIR for temporary Gaussian 03 files."

# Define node list

cat $PBS_NODEFILE|uniq > $GAUSS_SCRDIR/tsnet.nodes

NODE_NUM=`cat $PBS_NODEFILE|uniq |wc -l`

NP=`cat $PBS_NODEFILE|wc -l`

nodes=`cat $PBS_NODEFILE |uniq| awk '{printf("%s,",$1)}'`

sharecpu=`expr $NP / $NODE_NUM`

G03_NODEFILE="$GAUSS_SCRDIR/tsnet.nodes"

GAUSS_LFLAGS=" -mp 2 -nodefile $G03_NODEFILE"

export GAUSS_LFLAGS

echo pbs nodefile:

cat $G03_NODEFILE

#Run a Gaussian command file, water03.com, redirecting output to a file, water03.log

cd $GAUSS_RUNDIR

echo "%NProcShared=$sharecpu" > ${jobname}.Input.${PBS_JOBID}

if [ $NODE_NUM -ne 1 ];

then

echo "%LindaWorker=$nodes" >> ${jobname}.Input.${PBS_JOBID}

fi

grep -v nproc $INFILE |grep -v NProcShared |grep -v LindaWorker >>${jobname}.Input.${PBS_JOBID}

echo "Starting Gaussian run at" `date`

if [ $NODE_NUM -eq 1 ];

then

time g03 < ${jobname}.Input.${PBS_JOBID} > $GAUSS_RUNDIR/${jobname}.log

fi

else

time g03l < ${jobname}.Input.${PBS_JOBID} > $GAUSS_RUNDIR/${jobname}.log

fi

#time g03 < ${jobname}.Input.${PBS_JOBID} > ${jobname}.log

echo "Finished Gaussian run at" `date`

PBS_O_WORKDIR=${ORIG_PBS_O_WORKDIR}

echo $PBS_O_WORKDIR

mv $GAUSS_RUNDIR/${jobname}.* $PBS_O_WORKDIR

mv $GAUSS_SCRDIR/*.chk $PBS_O_WORKDIR

echo "$GAUSS_SCRDIR"

rm -Rf $GAUSS_SCRDIR

4、测试安装是否成功
准备算例test397.com

在算例目录下修改g03.pbs，然后执行qsub g03.pbs -v file=test397.com

cd ~/g03test

qsub g03.pbs -v file=test397.com

5、其他

三、 注意事项
1、本文命令、代码和超链接采用斜体五号字表示
2、算例文件名必须有两部分组成，前缀+后缀，中间用 . 隔开
3、需要修改一下两个文件以适应linda并行时的配置情况：
#vi /data2/home/test/g03/linda7.2/opteron-linux-I8/bin/LindaLauncher

/mf/giovanni/static/g03/linda7.2/opteron-linux-I8/bin/cLindaLauncher

#vi /data2/home/test//g03/ntsnet

/mf/giovanni/static/g03/linda7.2/opteron-linux-I8/bin/true_ntsnet

4、在所以参与计算的节点根目录上增加/scratch/，并设置开放的权限
mkdir /scratch

chmod 777 /scratch

5、其他

四、 参考文献
1 量子化学计算程序包GAUSSIAN 王涛 上海超级计算中心 上海 201203 [email protected]

⑤ 为什么香港中文大学研发的人脸识别算法能够击败人类

LFW（Labeled faces in the wild[1]）是人脸识别研究领域比较有名的人脸图像集合，其图像采集自Yahoo! News，共13233幅图像，其中5749个人，其中1680人有两幅及以上的图像，4069人只有一幅图像；大多数图像都是由Viola-Jones人脸检测器得到之后，被裁剪为固定大小，有少量的人为地从false positive中得到[2]。所有图像均产生于现实场景（有别于实验室场景），具备自然的光线，表情，姿势和遮挡，且涉及人物多为公物人物，这将带来化妆，聚光灯等更加复杂的干扰因素。因此，在该数据集上验证的人脸识别算法，理论上更贴近现实应用，这也给研究人员带来巨大的挑战。

⑥ 为什么香港中文大学研发的人脸识别算法能够击败人类

1.全面超过人脸时的条件是实验室内部拍摄条件、正面姿态、正面光照。这种条件下的人脸识别错误率的进展大约是每3年下降10倍。FRVT2012中期结果中最好的单位(不出意外应该是日本的NEC公司)的错误率已经达到了我们06年系统的1%左右。而我手上的系统相比06年大约提升了十几倍，目前在中期结果中排名6-7名。
2.lfw数据库直接是从雅虎网上抓的照片。难度在业界属于顶尖。该库09年公布后至今没有难度更大的静态照片库出现。难度相当但数据量更大的库倒是有两三个。我们06年的系统跑lfw也就70+的水平。而我们实验室的最高水平(也是国内除face++外的最高水平)大约是92左右。大概相当于2012年底的state-of-the-art。
3.2014年的三个逆天结果，deepface的97.25%、face++的97.27%、gaussianface的98.52%，前两者都用了deep learning。第一个训练数据400万。第二个算法细节不明，但deeplearning向来吃样本，想来训练库也是百万量级。唯有gaussianface的训练库仅2万余。
4.arxiv和CVPR等顶会完全不矛盾。先发上来只是为了不让别人抢先。估计未来的顶刊顶会上很快会出现这个结果。
5.算法细节太过技术，难以在这里深入浅出，就不多介绍了。只提一篇paper。Blei的latent dirichlet allocation，2003年的jmlr，引用量近万。本文对人脸的贡献方式大概相当于lda对文档分类的贡献方式，懂行的人自然知道这句话的分量。当然lda珠玉在前，deep learning风头正劲，所以真正的历史地位，本文自然不可能赶上lda。但一篇正常pami的水准肯定是有的。

⑦ 哪里能搞到Gaussian09的源代码 source code

正对自己的平台，使用源码编译，执行效率比较高。
我对比测试的是G09A01 和G03D02 目前流出的D02是别同通过source code编译的，
D02计算速度明显快很多的...

⑧ 在matlab中混入噪声功率为4W的随机噪声的代码是什么

M=imread('dl011.jpg') %读取MATLAB中的名为cameraman的图像
subplot(3,3,1)
imshow(M) %显示原始图像
title('original')
P1=imnoise(M,'gaussian',0.02) %加入高斯躁声
subplot(3,3,2)
imshow(P1) %加入高斯躁声后显示图像
title('gaussian noise');
P2=imnoise(M,'salt & pepper',0.02) %加入椒盐躁声
subplot(3,3,3)
imshow(P2) %%加入椒盐躁声后显示图像
title('salt & pepper noise');
g=medfilt2(P1) %对高斯躁声中值滤波
subplot(3,3,5)
imshow(g)
title('medfilter gaussian')
h=medfilt2(P2) %对椒盐躁声中值滤波
subplot(3,3,6)
imshow(h)
title('medfilter salt & pepper noise')
l=[1 1 1 %对高斯躁声算术均值滤波
1 1 1
1 1 1];
l=l/9;
k=conv2(P1,l)
subplot(3,3,8)
imshow(k,[])
title('arithmeticfilter gaussian')
%对椒盐躁声算术均值滤波
d=conv2(P2,l)
subplot(3,3,9)
imshow(d,[])
title('arithmeticfilter salt & pepper noise')

⑨ 高新波的论文成果

Souleymane Balla-Arabé, X.-B. Gao. A Fast and Robust Level Set Method for Image Segmentation Using Fuzzy Clustering and Lattice Boltzmann Method. IEEE Trans. on Systems, Man and Cybernetics Part B: Cybernetics (IEEE TSMC B), 2013. In Press
K.-B. Zhang, X.-B. Gao, et al. Learning Local and Non-Local Priors for Single Image Super-resolution. IEEE Trans. Image Processing (IEEE TIP), Vol.21, No,11, pp.4544-4556, 2012.
L.-L. An, X.-B. Gao, et al. Robust Reversible Watermarking via Clustering and Enhanced Pixel-wise Masking. IEEE Trans. Image Processing (IEEE TIP), Vol.21, No.8, pp.3589-3611, 2012.
Y. Su, X.-B. Gao, et al. Multivariate Multi-linear Regression. IEEE Trans. on Systems, Man and Cybernetics Part B: Cybernetics (IEEE SMC B), Vol.42, No.6, pp.1560-1573, Dec. 2012.
X.-B. Gao, K. Zhang, et al. Image Super-Resolution with Sparse Neighbor em[ant]bedding. IEEE Trans. Image Processing (IEEE TIP), Vol.21, No.7, pp.3149-3205, 2012.
Y. Su, Y. Fu, X.-B Gao, Q. Tian. Discriminant Learning through Multiple Principal Angles for Visual Recognition. IEEE Trans. on Image Processing (IEEE TIP), Vol.21, No.3, pp.1381-1390, 2012.
X.-B Gao, K.-B. Zhang, et al. Joint Learning for Single Image Super-resolution via Coupled Constraint. IEEE Trans. on Image Processing (IEEE TIP), Vol.21, No.2, pp.469-490, 2012.
C.-N. Tian, G. Fan, X.-B. Gao, Q. Tian. Multi-view Face Recognition: From TensorFace to V-TensorFace and K-TensorFace. IEEE Trans. on Systems, Man and Cybernetics Part B: Cybernetics (IEEE TSMC B). Vol.42, No.2, pp.320-333, 2012.
X.-B. Gao, N.-N. Wang, et al. Face Sketch-Photo Synthesis and Retrieval Using Sparse Representation. IEEE Trans. on Circuits Systems for Video technology (IEEE TCSVT), Vol.22, No.8, pp.1213-1226, 2012
X.-B. Gao, X. M. Wang, et al. Supervised Gaussian Process Latent Variable Model for Dimensionality Rection. IEEE Trans. on System, Man, and Cybernetics, Part B: Cybernetics (IEEE TSMC B), Vol.41, No.2, pp.518 525, April 2011.
X.-B. Gao, B. Wang, et al. A Relay Level Set Method for Automatic Image Segmentation. IEEE Trans. on System, Man, and Cybernetics, Part B: Cybernetics (IEEE TSMC B), Vol.41, No.2, pp.42 434, April 2011.
X.-B. Gao, L. L. An, et al. Lossless Data em[ant]bedding Using Generalized Statistical Quantity Histogram. IEEE Trans. on Circuits Systems for Video Technology (IEEE TCSVT), Vol.21, No.8, pp.1061 1070, 2011.
K.-B. Zhang, X.-B. Gao, et al. Partially Supervised Neighbor em[ant]bedding for Example based Image Super resolution. IEEE Journal of Selected Topics in Signal Processing, Vol.5, No.5, pp.230 239, 2011.
X.-B. Gao, J. Chen, et al. Multi sensor Centralized Fusion without Measurement Noise Covariance by Variational Bayesian Approximation. IEEE Trans. on Aerospace and Electronic Systems (IEEE TAES), Vol.47, No.1, pp.718 722, 2011.
X.-B. Gao, Q. Wang, et al. Zernike Moment based Image Super Resolution. IEEE Trans. on Image Processing (IEEE TIP), Vol. 20, No.10, pp.2738 2747, 2011.
X.-B. Gao, C. Deng, et al. Geometric Distortion Insensitive Image Watermarking in Affine Covariant Regions. IEEE Trans. on System, Man and Cybernetics, Part C: Applications and Reviews (IEEE TSMC C), Vol.40, No.3, pp.278 286, 2010.
X.-B. Gao, Y. Su, et al. A Review of Active Appearance Models. IEEE Trans. on System, Man, and Cybernetics, Part C: Applications and Reviews (IEEE TSMC C), Vol.40, No.2, pp.145 158, 2010.
X.-B. Gao, Y. Wang, et al. On Combining Morphological Component Analysis and Concentric Morphology Model for Mammographic Mass Detection. IEEE Trans. on Information Technology in Biomedicine (IEEE TITB), Vol.14, No.2, pp.266 273, 2010.
B. Wang, X.-B. Gao, et al. A Unified Tensor Level Set for Image Segmentation. IEEE Trans. on System, Man, and Cybernetics, Part B: Cybernetics (IEEE TSMC B), Vol. 40, No.3, pp.857 867, 2010.
X.-B. Wang, Z. Li, P. C. Xu, Y. Y. Xu, X.-B. Gao. Spectrum Sharing in Cognitive Radio Networks: An Auction based Approach. IEEE Trans. on Systems, Man and Cybernetics Part B: Cybernetics (IEEE TSMC B), Vol.40, No.3, pp.587 596, June 2010.
C. H. Hu, X. B. Wang, Z. C. Yang, J. F. Zhang, Y. Y. Xu, X.-B. Gao. A Geometry Study on the Capacity of Wireless Networks via Percolation. IEEE Trans. on Communications (IEEE TC), Vol.58, No.10, pp.2916 2925, 2010.
X.-B. Gao, W. Lu, et al. Image Quality Assessment Based on Multiscale Geometric Analysis. IEEE Trans. on Image Processing (IEEE TIP), Vol.18, No.7, pp.1409 1423, 2009.
D. Tao, X. Li, W. Lu, X.-B. Gao. Reced reference IQA in Contourlet Domain. IEEE Trans. on System, Man, and Cybernetics, Part B: Cybernetics (IEEE TSMC B), Vol.39, No.6, pp.1623 1627, 2009.
X.-B. Gao, J. Zhong, J. Li, C. Tian. Face Sketch Synthesis Algorithm Based on E HMM and Selective Ensemble. IEEE Trans. on Circuits and Systems for Video Technology (IEEE TCSVT), Vol.18, No.4, pp.487 496, 2008.
X.-B. Gao and X. Tang. Unsupervised Video Shot Segmentation and Model free Anchorperson Detection for News Video Story Parsing. IEEE Trans. on Circuits Systems for Video Technology (IEEE TCSVT), Vol. 12, no. 9, pp.765 776, 2002.
X. Tang, X.-B. Gao, J. Z. Liu and H. Zhang. A Spatial temporal Approach for Video Caption Detection and Recognition. IEEE Trans. on Neural Networks (IEEE TNN), Vol.13, No. 4, pp. 961 971, 2002.
Z. X. Niu, X.-B. Gao, Q. Tian. Real World Trajectory Extraction for Attack Pattern Analysis in Soccer Video. Pattern Recognition, Vol.45, No.5, pp.1937 1947, 2012.
X.-B. Gao, X. Wang, et al. Transfer Latent Variable Model Learning Based on Divergence Analysis. Pattern Recognition (Elsevier), Vol.44, No.10 11, pp.2358 2366, 2011.
Y. Wang, D. Tao, X.-B. Gao. Feature em[ant]bedded vector valued contour based level set method with relaxed shape constraint for mammographic mass segmentation. Pattern Recognition (Elsevier), Vol.44, No.9, pp.1903 1915, 2011.
X.-B. Gao, B. Xiao, et al. Image categorization: graph edit distance + edge direction histogram. Pattern Recognition (Elsevier). Vol.41, No.10, pp.3179 3191, October, 2008.
X.-B. Gao, et al. Shot based Video Retrieval with Optical Flow Tensor and HMM. Pattern Recognition Letters (Elsevier), Vol.30, No.2, pp.140 147, 2009.
X. Chen, X.-B. Gao*, et al. 3D Reconstruction of Light Flux Distribution on Arbitrary Surfaces from 2D Multi photographic Images. Optics Express, Vol,18, No.19, pp.19876 19893, 2010.
X. Chen, X.-B. Gao, et al. A study of photon propagation in free space based on hybrid radiosity radiance theorem. Optics Express, Vol.17, No.18, pp.16266 16280, 2009
X.-B. Gao, R. Fu, et al. Image Segmentation for Aurora Index Extraction. Computer Vision and Image Understanding, Vol.115, No.3, pp.390 402, 2011.
X.-B. Gao, Y. M. Yang, et al. Discriminative optical flow tensor for video semantic analysis. Computer Vision and Image Understanding, Vol.113, No. 3, pp.372 383, 2009.
K. Zhang, X.-B. Gao, et al. Multi scale Dictionary for Single Image Super resolution. Proceedings of Computer Vision and Pattern Recognition (CVPR2012), Providence, Rhode Island, 16 21 June, 2012, USA.
Z. Niu, G. Hua, X.-B. Gao and Q. Tian. Context Aware Topic Model for Scene Recognition. Proceedings of Computer Vision and Pattern Recognition (CVPR2012), Providence, Rhode Island, 16 21 June, 2012, USA.
L. H. He, D. Tao, X. Li, X.-B. Gao. Sparse Representation for Blind Image Quality Assessment. Proceedings of Computer Vision and Pattern Recognition (CVPR2012), Providence, Rhode Island, 16 21 June, 2012, USA.
Z. X. Niu, G. Hua, X.-B. Gao, Q. Tian. Spatial DiscLDA for Visual Recognition. Proceedings of Computer Vision and Pattern Recognition (CVPR2011), 21 23 June, 2011, Colorado, USA.
Z. X. Niu, Q. Tian, X.-B. Gao. Real World Trajectory Extraction for Attack Pattern Analysis in Soccer Video. Proceedings of the ACM International Conference on Multimedia (ACM MM2010), pp.635 638, 25 29 October 2010, Firenze, Italy.
W. Ning, J. Li, J. Li, X.-B. Gao. 3D Medical Image Processing and Analyzing System. Demo session of Asian Conference on Computer Vision (ACCV2009), Xian, China, 2009.
其它核心期刊及国际会议论文200余篇，其中SCI检索100余篇，EI检索200余篇。