I am searching for any matlab code that reads the skeleton text files from the dataset (MSR Daily Activity 3D). I can't figure how to understand how the files are written and what they represent ? Also, don't know how to parse them to extract the features.
this is the matlab code which read the depth sequences.
in each loop of 'for ei = 1:2',it read all the depth data to depth(a 3D matrix) from a bin file.
clear;close all;clc;
binPath = 'MSR Daily Activity 3D dataset\Depth';
for ai = 1:16
for si = 1:10
for ei = 1:2
%%%%%%%%%%%%%
%%%%%%%%%%%%%
[acsr,susr,exsr]=getsr(ai,si,ei);
%%%%%% getsr(ai,si,ei) convert ai,si,ei to double bits
%%%%%% for example, if ai=3, acsr is 03
%%%%%%%%%%%
binfile = [binPath,'\a',acsr,'_s',susr,'_e',exsr,'_depth.bin'];
if ~exist(binfile,'file');
disp('error');
continue;
end;
disp(binfile);
fileread = fopen(binfile);
if fileread<0
disp('no such file.');
return;
end
header = fread(fileread,3,'uint=>uint');
nnof = header(1); ncols = header(2); nrows = header(3);
depths = zeros(ncols, nrows, nnof);
for f = 1:nnof
frame = zeros( ncols, nrows);
for row = 1:nrows
tempRow = fread(fileread, ncols, 'uint=>uint');
tempRowID = fread(fileread, ncols, 'uint8');%%%%%
frame(:,row) = tempRow;
end
depth(:,:,f) = frame;
clear tempRow tempRowID;
end
fclose(fileread);
end
end
end
The site http://research.microsoft.com/en-us/um/people/zliu/ActionRecoRsrc/ tells you exactly how they're organised. They also provide some C++ example loaders.
"The format of the skeleton file is as follows. The first integer is the number of frames. The second integer is the number of joints which is always 20. For each frame, the first integer is the number of rows. This integer is 40 when there is exactly one skeleton being detected in this frame. It is zero when no skeleton is detected. It is 80 when two skeletons are detected (in that case which is rare, we simply use the first skeleton in our experiments). For most of the frames, the number of rows is 40. Each joint corresponds to two rows. The first row is its real world coordinates (x,y,z) and the second row is its screen coordinates plus depth (u, v, depth) where u and v are normalized to be within [0,1]. For each row, the integer at the end is supposed to be the confidence value, but it is not useful."
Hope this helps.
Related
I am using MATLAB version R2011a, a friend of mine is using R2014b which contains the function "Flip", which flips the order of elements, this function is vital to our program that compares Matrix'es.
My problem is R2011a does not have this function, it has fliplr,flipud and flipdim. I have tried using fliplr and then flipud to try and recreate the same function but eventually it doesn't work since i'm using the function corr which requires using that it's two arguments be the same dimensions.
I need advise on how to create the flip function that is available on R2014b.
The function that is problematic:
%This function gets the DNA signiture with the relative freq of each perm at
%the refernce text, the DNA signiture with the relative freq of each perm at
%the compare text, and the MaxPerm, and return the relative editor distance
%between the 2 texts.
function [distance]=EditorDistance2 (RefDNAWithFreq,CmpDNAWithFreq,MaxPerm)
if MaxPerm>2
MaxPerm=2;
end
str='Editor Distance compare begun';
disp(str);
distance=[];
for PermLength=1:MaxPerm
freq=sum(0:PermLength);
PermInitial=freq+1;
permEnd=freq+PermLength;
%create an ordered matrix of all the perms with length "PermLength"
%in the ref text
CurRefPerms=RefDNAWithFreq(:,freq:permEnd);
OrderedRefCurPerms=sortrows(CurRefPerms);
OrderedRefCurPerms=flip(OrderedRefCurPerms);
OrderedRefCurPerms(:,1)=[];
OrderedRefCurPerms=ZeroCutter(OrderedRefCurPerms);
%create an ordered matrix of all the perms with length "PermLength"
%in the cmp text
CurcmpPerms=CmpDNAWithFreq(:,freq:permEnd);
OrderedCmpCurPerms=sortrows(CurcmpPerms);
OrderedCmpCurPerms=flip(OrderedCmpCurPerms);
OrderedCmpCurPerms(:,1)=[];
OrderedCmpCurPerms=ZeroCutter(OrderedCmpCurPerms);
len1=size(OrderedRefCurPerms,1);
len2=size(OrderedCmpCurPerms,1);
edit=1;
matrix=zeros(len2,len1);
%initiate first row of the first stirng
for i=2:len1
matrix(1,i)=matrix(1,i-1)+1;
end
%initiate first column of the second stirng
for i=2:len2
matrix(i,1)=matrix(i-1,1)+1;
end
%start algoritem
for i=2:len2
for j=2:len1
if OrderedRefCurPerms(j-1,:)==OrderedCmpCurPerms(i-1,:)
edit=0;
end
if (i>2 & j>2 & OrderedRefCurPerms(j-1,:)==OrderedCmpCurPerms(i-2,:) & RefDNAWithFreq(j-2)==CmpDNAWithFreq(i-1) )
matrix(i,j)= min([matrix(i-1,j)+1,... deletion
matrix(i,j-1)+1,... insertion
matrix(i-2,j-2)+1,... substitution
matrix(i-1,j-1)+edit... transposition
]);
else
matrix(i,j) = min([matrix(i-1,j)+1,... deletion
matrix(i,j-1)+1,... insertion
matrix(i-1,j-1)+edit... substitution
]);
end
edit=1;
end
end
%The Distance is the last elment of the matrix.
if i~=1
tempdistance = matrix( floor( len2 / 3 ) , floor( len1 / 3 ) );
tempdistance=tempdistance/floor(len2/3);
else
tempdistance = matrix( len2,len1 );
tempdistance= tempdistance/len2;
end
tempdistance=1-tempdistance;
distance=[distance tempdistance];
end
end
I will further explain myself, the function which I am trying to use is A=flip(A)
The function that causes me problems is this one
%This function gets the DNA signiture with the relative freq of each perm at
%the refernce text, the DNA signiture with the relative freq of each perm at
%the compare text, and the MaxPerm, and return the corralation between the 2 texts.
function [Corvector]=CorrelationCompare(RefDNAWithFreq,CmpDNAWithFreq,MaxPerm)
str='corraltion compare begun';
disp(str);
%this vector will contain the corralation between the freqs of
%each perms vector(each length)
Corvector=[];
for PermLength=1:MaxPerm
freq=sum(0:PermLength);
PermInitial=freq+1;
permEnd=freq+PermLength;
%Cor is correlation between the 2 texts
refPerms=RefDNAWithFreq(:,freq);
cmpPerms=CmpDNAWithFreq(:,freq);
refPerms=ZeroCutter(refPerms);
cmpPerms=ZeroCutter(cmpPerms);
tempCor=corr(refPerms,cmpPerms);
Corvector =[Corvector tempCor];
% making a graph of the perms, and the relative freq of the texts.
x=ZeroCutter ( RefDNAWithFreq(:,PermInitial:permEnd) );
y1=refPerms;
y2=cmpPerms;
xchars=char(x);
Xcols=size(x,1);
o=ones(Xcols,1);
xco=mat2cell(xchars,o,PermLength);
xaxis=(1:Xcols);
figure
stem(xaxis,y1,'r');
hold
stem(xaxis,y2,'g');
set(gca,'XTick',xaxis)
set(gca,'XTickLabel',xco,'fontname','david');
xlabel('Perms');
ylabel('Perm frequency');
TitleOfGraph=sprintf('comapre between reference text to the compared, %d letters perm\n correlation=%f',PermLength,Corvector(PermLength));
legend('reference','compared');
title(TitleOfGraph);
end
end
The Error that I recieve when using a diffrent flip command is
??? Error using ==> corr at 102
X and Y must have the same number of rows.
Error in ==> CorrelationCompare at 27
tempCor=corr(refPerms,cmpPerms);
I apologize for the long codes but it's hard to explain it all since it's a big project and a lot of it was done by my partner
This should work for you -
function out = flip_hacked(A,dim)
%// Get an array of all possible dimensions
dims = 1:ndims(A);
%// Interchange first dimension and dim
dims(dim) = 1;
dims(1) = dim;
A1 = permute(A,[dims]);
%// Reshape A1 into a 2D matrix and then flip along the first dimension,
%// which would correspond to the flipping along dim and then interchange dim
%// and first dim again to keep the size of data same as input and elements
%// being flipped along dim for the desired output
A2 = reshape(A1,size(A1,1),[]);
out = permute(reshape(A2(end:-1:1,:),size(A1)),dims);
return;
It follows the same syntax as the official flip function that's stated in the official documentation as follows -
B = flip(A,dim) reverses the order of the elements in A along
dimension dim. For example, if A is a matrix, then flip(A,1) reverses
the elements in each column, and flip(A,2) reverses the elements in
each row.
In addition to the generic solution provided by Divakar you could simply use:
flip = #(A) A(end:-1:1, :);
A = flip(A);
To reverse the elements in each column of a matrix A. Even simpler:
A = A(end:-1:1, :);
My code works in the following manner:
1.First, it obtains several images from the training set
2.After loading these images, we find the normalized faces,mean face and perform several calculation.
3.Next, we ask for the name of an image we want to recognize
4.We then project the input image into the eigenspace, and based on the difference from the eigenfaces we make a decision.
5.Depending on eigen weight vector for each input image we make clusters using kmeans command.
Source code i tried:
clear all
close all
clc
% number of images on your training set.
M=1200;
%Chosen std and mean.
%It can be any number that it is close to the std and mean of most of the images.
um=60;
ustd=32;
%read and show images(bmp);
S=[]; %img matrix
for i=1:M
str=strcat(int2str(i),'.jpg'); %concatenates two strings that form the name of the image
eval('img=imread(str);');
[irow icol d]=size(img); % get the number of rows (N1) and columns (N2)
temp=reshape(permute(img,[2,1,3]),[irow*icol,d]); %creates a (N1*N2)x1 matrix
S=[S temp]; %X is a N1*N2xM matrix after finishing the sequence
%this is our S
end
%Here we change the mean and std of all images. We normalize all images.
%This is done to reduce the error due to lighting conditions.
for i=1:size(S,2)
temp=double(S(:,i));
m=mean(temp);
st=std(temp);
S(:,i)=(temp-m)*ustd/st+um;
end
%show normalized images
for i=1:M
str=strcat(int2str(i),'.jpg');
img=reshape(S(:,i),icol,irow);
img=img';
end
%mean image;
m=mean(S,2); %obtains the mean of each row instead of each column
tmimg=uint8(m); %converts to unsigned 8-bit integer. Values range from 0 to 255
img=reshape(tmimg,icol,irow); %takes the N1*N2x1 vector and creates a N2xN1 matrix
img=img'; %creates a N1xN2 matrix by transposing the image.
% Change image for manipulation
dbx=[]; % A matrix
for i=1:M
temp=double(S(:,i));
dbx=[dbx temp];
end
%Covariance matrix C=A'A, L=AA'
A=dbx';
L=A*A';
% vv are the eigenvector for L
% dd are the eigenvalue for both L=dbx'*dbx and C=dbx*dbx';
[vv dd]=eig(L);
% Sort and eliminate those whose eigenvalue is zero
v=[];
d=[];
for i=1:size(vv,2)
if(dd(i,i)>1e-4)
v=[v vv(:,i)];
d=[d dd(i,i)];
end
end
%sort, will return an ascending sequence
[B index]=sort(d);
ind=zeros(size(index));
dtemp=zeros(size(index));
vtemp=zeros(size(v));
len=length(index);
for i=1:len
dtemp(i)=B(len+1-i);
ind(i)=len+1-index(i);
vtemp(:,ind(i))=v(:,i);
end
d=dtemp;
v=vtemp;
%Normalization of eigenvectors
for i=1:size(v,2) %access each column
kk=v(:,i);
temp=sqrt(sum(kk.^2));
v(:,i)=v(:,i)./temp;
end
%Eigenvectors of C matrix
u=[];
for i=1:size(v,2)
temp=sqrt(d(i));
u=[u (dbx*v(:,i))./temp];
end
%Normalization of eigenvectors
for i=1:size(u,2)
kk=u(:,i);
temp=sqrt(sum(kk.^2));
u(:,i)=u(:,i)./temp;
end
% show eigenfaces;
for i=1:size(u,2)
img=reshape(u(:,i),icol,irow);
img=img';
img=histeq(img,255);
end
% Find the weight of each face in the training set.
omega = [];
for h=1:size(dbx,2)
WW=[];
for i=1:size(u,2)
t = u(:,i)';
WeightOfImage = dot(t,dbx(:,h)');
WW = [WW; WeightOfImage];
end
omega = [omega WW];
end
% Acquire new image
% Note: the input image must have a bmp or jpg extension.
% It should have the same size as the ones in your training set.
% It should be placed on your desktop
ed_min=[];
srcFiles = dir('G:\newdatabase\*.jpg'); % the folder in which ur images exists
for b = 1 : length(srcFiles)
filename = strcat('G:\newdatabase\',srcFiles(b).name);
Imgdata = imread(filename);
InputImage=Imgdata;
InImage=reshape(permute((double(InputImage)),[2,1,3]),[irow*icol,1]);
temp=InImage;
me=mean(temp);
st=std(temp);
temp=(temp-me)*ustd/st+um;
NormImage = temp;
Difference = temp-m;
p = [];
aa=size(u,2);
for i = 1:aa
pare = dot(NormImage,u(:,i));
p = [p; pare];
end
InImWeight = [];
for i=1:size(u,2)
t = u(:,i)';
WeightOfInputImage = dot(t,Difference');
InImWeight = [InImWeight; WeightOfInputImage];
end
noe=numel(InImWeight);
% Find Euclidean distance
e=[];
for i=1:size(omega,2)
q = omega(:,i);
DiffWeight = InImWeight-q;
mag = norm(DiffWeight);
e = [e mag];
end
ed_min=[ed_min MinimumValue];
theta=6.0e+03;
%disp(e)
z(b,:)=InImWeight;
end
IDX = kmeans(z,5);
clustercount=accumarray(IDX, ones(size(IDX)));
disp(clustercount);
Running time for 100 images:Elapsed time is 103.947573 seconds.
QUESTIONS:
1.It is working fine for M=50(i.e Training set contains 50 images) but not for M=1200(i.e Training set contains 1200 images).It is not showing any error.There is no output.I waited for 10 min still there is no output.What is the problem?Where i was wrong?
To answer your second question, you can simply 'save' any generated variable as a .mat file in your working directory (Current folder) which can be accessed later. So in your code, if the 'training eigenfaces' is given by the variable 'u', you can use the following:
save('eigenface.mat','u')
This creates a .mat file with the name eigenface.mat which contains the variable 'u', the eigenfaces. Note that this variable is saved in your Current Folder.
In a later instance when you are trying out with your test data, you can simply 'load' this variable:
load('eigenface.mat')
This automatically loads 'u' into your workspace.
You can also save additional variables in the same .mat file if necessary
save('eigenface.mat','u','v'...)
The answer to the first question is that the code is simply not done running yet. Vectorizing the code (instead of using a for loop), as suggested in the comment section above can improve the speed significantly.
[EDIT]
Since the images are not very big, the code is not significantly slowed down by the first for loop. You can improve the performance of the rest of the code by vectorizing the code. Vectorized operations are faster than for-loops. This link might be useful in understanding vectorization:
http://www.mathworks.com/help/matlab/matlab_prog/vectorization.html
For example, the second for-loop can be replaced by the following vectorized form, as suggested in the comments
tempS = double(S);
meanS = mean(S,1);
stdS = std(S,0,1);
S = (tempS - meanS) ./ stdS;
Use MATLAB's timer functions tic and toc for finding how long the first for-loop alone is taking to execute. Add tic before the for loop and toc after it. If the time taken for 50 images is about 104 seconds, then it would be significantly more for 1200 images.
I have a data set(binary file) which i want to read only the first half of X (and corresponding Y) data which is saved to 4D matrix:
for i = 1:vols
for j = 1:cols
XY(i,:,:,j) = fread(fid,[X Y],'int16');
end
end
How do I modify the above loop so only the first e.g. 10 X data (and corresponding Y) is read in for each vols and cols?
thanks
You will need to implement reading for each vols and cols in following order:
read part of Y for the first input X, than skip rest of this line, read part of Y for the second input X, etc.
After reading of requested number of X lines, you will need to skip rest of matrix before read next (vols, cols) pair.
To skip part of matrix you can use fseek function.
Let X_count and Y_cound are dimensions of submatrix; X_total and Y_total are dimension of total matrix. You need something like following:
for i = 1:vols
for j = 1:cols
for k=1:X_count
XY(i,k,:,j) = fread(fid,Y_count,'int16');
fseek(fid,(Y_total-Y_count)*2,'cof');
end
fseek(fid,(X_total-X_count)*Y_total*2,'cof');
end
end
Okay so this sounds easy but no matter how many times I have tried I still cannot get it to plot correctly. I need only 3 lines on the same graph however still have an issue with it.
iO = 2.0e-6;
k = 1.38e-23;
q = 1.602e-19;
for temp_f = [75 100 125]
T = ((5/9)*temp_f-32)+273.15;
vd = -1.0:0.01:0.6;
Id = iO*(exp((q*vd)/(k*T))-1);
plot(vd,Id,'r',vd,Id,'y',vd,Id,'g');
legend('amps at 75 F', 'amps at 100 F','amps at 125 F');
end;
ylabel('Amps');
xlabel('Volts');
title('Current through diode');
Now I know the plot function that is currently in their isn't working and that some kind of variable needs setup like (vd,Id1,'r',vd,Id2,'y',vd,Id3,'g'); however I really can't grasp the concept of changing it and am seeking help.
You can use the "hold on" function to make it so each plot command plots on the same window as the last.
It would be better to skip the for loop and just do this all in one step though.
iO = 2.0e-6;
k = 1.38e-23;
q = 1.602e-19;
temp_f = [75 100 125];
T = ((5/9)*temp_f-32)+273.15;
vd = -1.0:0.01:0.6;
% Convert this 1xlength(vd) vector to a 3xlength(vd) vector by copying it down two rows.
vd = repmat(vd,3,1);
% Convert this 1x3 array to a 3x1 array.
T=T';
% and then copy it accross to length(vd) so each row is all the same value from the original T
T=repmat(T,1,length(vd));
%Now we can calculate Id all at once.
Id = iO*(exp((q*vd)./(k*T))-1);
%Then plot each row of the Id matrix as a seperate line. Id(1,:) means 1st row, all columns.
plot(vd,Id(1,:),'r',vd,Id(2,:),'y',vd,Id(3,:),'g');
ylabel('Amps');
xlabel('Volts');
title('Current through diode');
And that should get what you want.
In Matlab, I want to make a seqlogo plot of an amino acid sequence profile. But instead of scaling the heights of the plot columns by entropy, I want all the columns to be the same height.
I'm in the process of modifying the code from the answers to this question, but I wonder if there is a parameter to seqlogo or some other function that I have missed that will make the column heights uniform.
Alternatively, is there a statistical transformation I can apply to the sequence profile to hack the desired output? (column heights uniform, height of each letter linearly proportion to
its probability in the seqprofile)
Probably the easiest way around this problem is to directly modify the code for the Bioinformatics Toolbox function SEQLOGO (if possible). In R2010b, you can do:
edit seqlogo
And the code for the function will be shown in the editor. Next, find the following lines (lines 267-284) and either comment them out or remove them entirely:
S_before = log2(nSymbols);
freqM(freqM == 0) = 1; % log2(1) = 0
% The uncertainty after the input at each position
S_after = -sum(log2(freqM).*freqM, 1);
if corrError
% The number of sequences correction factor
e_corr = (nSymbols -1)/(2* log(2) * numSeq);
R = S_before - (S_after + e_corr);
else
R = S_before - S_after;
end
nPos = (endPos - startPos) + 1;
for i =1:nPos
wtM(:, i) = wtM(:, i) * R(i);
end
Then put this line in their place:
wtM = bsxfun(#times,wtM,log2(nSymbols)./sum(wtM));
You will probably want to save the file under a new name, like seqlogo_norm.m, so you can still use the original unmodified SEQLOGO function. Now you can create sequence profile plots with all the columns normalized to the same height. For example:
S = {'LSGGQRQRVAIARALAL',... %# Sample amino acid sequence
'LSGGEKQRVAIARALMN',...
'LSGGQIQRVLLARALAA',...
'LSGGERRRLEIACVLAL',...
'FSGGEKKKNELWQMLAL',...
'LSGGERRRLEIACVLAL'};
seqlogo_norm(S,'alphabet','aa'); %# Use the modified SEQLOGO function
OLD ANSWER:
I'm not sure how to transform the sequence profile information to get the desired output from the Bioinformatics Toolbox function SEQLOGO, but I can show you how to modify the alternative seqlogo_new.m that I wrote for my answer to the related question you linked to. If you change the line that initializes bitValues from this:
bitValues = W{2};
to this:
bitValues = bsxfun(#rdivide,W{2},sum(W{2}));
Then you should get each column scaled to a height of 1. For example:
S = {'ATTATAGCAAACTA',... %# Sample sequence
'AACATGCCAAAGTA',...
'ATCATGCAAAAGGA'};
seqlogo_new(S); %# After applying the above modification
For now, my workaround is to generate a bunch of fake sequences that match the sequence profile, then feed those sequences to http://weblogo.berkeley.edu/logo.cgi . Here is the code to make the fake sequences:
function flatFakeSeqsFromPwm(pwm, letterOrder, nSeqsToGen, outFilename)
%translates a pwm into a bunch of fake seqs with the same probabilities
%for use with http://weblogo.berkeley.edu/
%pwm should be a 4xn or a 20xn position weight matrix. Each col must sum to 1
%letterOrder = e.g. 'ARNDCQEGHILKMFPSTWYV' for my data
%nSeqsToGen should be >= the # of pixels tall you plan to make your chart
[height windowWidth] = size(pwm);
assert(height == length(letterOrder));
assert(isequal(abs(1-sum(pwm)) < 1.0e-10, ones(1, windowWidth))); %assert all cols of pwm sum to 1.0
fd = fopen(outFilename, 'w');
for i = 0:nSeqsToGen-1
for seqPos = 1:windowWidth
acc = 0; %accumulator
idx = 0;
while i/nSeqsToGen >= acc
idx = idx + 1;
acc = acc + pwm(idx, seqPos);
end
fprintf(fd, '%s', letterOrder(idx));
end
fprintf(fd, '\n');
end
fclose(fd);
end