How can I load data from a folder in a variable using a for loop?

Question

0 个投票

Hi there, I am new to MATLAB and programming in general. I'm using code from Github for a research project in which participants draw where they experiencing bodily sensations (e.g., increasing heart rate) in the context of a specific emotion.

I want to load the mouse movements of my subjects into the variable "data". While doing so, it seems I get stuck in a for-while loop. It seems that the variable "data" is not created (at least it does not show up in my Workspace). I do not receive any error messages.

I don't know what could go wrong and where to look for solutions. Am I missing something?

I have copy pasted the code below.

% get a list of subjects
basepath='demo_subjects'; % folder where subjects are
subjects=dir([basepath '\D:\MATLAB\embody-master\embody_presentation_jan2023\*']);
% the base image used for painting (in our case only one sided since we
% subtract values)
base=uint8(imread('base.png'));
base2=base(10:531,33:203,:); % single image base
labels={'Neutral'
'Fear'
'Anger'
'Disgust'
'Sadness'
'Happiness'
'Surprise'
'Anxiety'
'Love'
'Depression'
'Contempt'
'Pride'
'Shame'
'Jealousy'
};
mask=imread('mask.png');
     % for each subject, load data
for s=1:length(subjects)
   
    % skip dot and dotdot folders
    if(strcmp(subjects(s).name(1),'.')) 
        continue; end 

I think Data Loading part is where it goes wrong. For completeness I have pasted the entire for-while loop (I feel like it's quite a long loop?).

    %% Data loading
    % let's load the subject's answers into a variable 
    data=load_subj([basepath '/' subjects(s).name],2);
    NC=length(data); % number of conditions 
    
    %% Painting reconstruction
    % 'data' now contains all mouse movements. What we need are the mouse
    % locations while the button was pressed (i.e. during painting)
    % Furthermore, the painting tool has a brush size. We recreate that
    % using image filter
    
for n=1:NC
        T=length(data(n).paint(:,2)); % number of mouse locations
        over=zeros(size(base,1),size(base,2)); % empty matrix to reconstruct painting
        for t=1:T
            y=ceil(data(n).paint(t,3)+1);
            x=ceil(data(n).paint(t,2)+1);
            if(x<=0) x=1; end
            if(y<=0) y=1; end
            if(x>=900) x=900; end % hardcoded for our experiment, you need to change it if you changed layout
            if(y>=600) y=600; end % hardcoded for our experiment, you need to change it if you changed layout
            over(y,x)=over(y,x)+1;
        end
        % Simulate brush size with a gaussian disk
        h=fspecial('gaussian',[15 15],5);
        over=imfilter(over,h);
        % we subtract left part minus right part of painted area
        % values are hard-coded to our web layout
        over2=over(10:531,33:203,:)-over(10:531,696:866,:);
        resmat(:,:,n)=over2;
    end
    %% store result (commented)
    save(['preprocessed' subjects(s).name '_preprocessed.mat'],'resmat')
    
    %% visualize subject's data
    M=max(abs(resmat(:))); % max range for colorbar
    NumCol=64;
    hotmap=hot(NumCol);
    coldmap=flipud([hotmap(:,3) hotmap(:,2) hotmap(:,1) ]);
    hotcoldmap=[
        coldmap
        hotmap
    ]
    % note that - for statistical maps - if you want to hide non
    % significant values and show them as black, you need to tweak the
    % colormap so that you have more rows of black between around the
    % non-significant interval. 
    % As an example if we had a threshold, uncomment the below
    if(0)
        th=0.2*M; % just an example threshold, since this is not a statistical map 
        non_sig=round(th/M*NumCol); % proportion of non significant colors
        hotmap=hot(NumCol-non_sig);
        coldmap=flipud([hotmap(:,3) hotmap(:,2) hotmap(:,1) ]);
        hotcoldmap=[
            coldmap
            zeros(2*non_sig,3);
            hotmap
        ]
    end
        
	% visualize all responses for each subject into a grid of numcolumns
	plotcols = 7; %set as desired
    plotrows = ceil((NC+1)/plotcols); % number of rows is equal to number of conditions+1 (for the colorbar)
    for n=1:NC
        figure(s)
        subplot(plotrows,plotcols,n)
        imagesc(base2);
        axis('off');
        set(gcf,'Color',[1 1 1]);
        hold on;
        over2=resmat(:,:,n);
        fh=imagesc(over2,[-M,M]);
        axis('off');
        axis equal
        colormap(hotcoldmap);
        set(fh,'AlphaData',mask)
        title(labels(n),'FontSize',10)
        if(n==NC) 
            subplot(plotrows,plotcols,n+1)
            fh=imagesc(ones(size(base2)),[-M,M]);
            axis('off');
            colorbar;
            % save a screenshot, useful for quality control (commented)
            saveas(gcf,[subjects(s).name '.png'])
        end
    end
    end 

The code uses a function "load_subj" which, I think, reads the list of conditions that the subject was presented (these are stored in a txt file) and "connects" it to the correct mouse movement data (stored in several csv files in the subject folder).

    function data=load_subj(folder,option)
    list=csvread([folder 'presentation.txt']);
    N=length(list);
    if(option==0)
        disp('here')
        for n=0:N-1
            file=[folder '/' num2str(n) '.csv'];
            fid = fopen(file);
            line=textscan(fid,'%s','CollectOutput',1,'Delimiter',';');
            data(:,n+1)=line{1};
        end
        data=data';
    end
    if(option==1)
        disp('option 1 not implemented')
    end
    if(option>=2)
        for n=0:N-1
            file=[folder '\' num2str(n) '.csv'];
            
            line=dlmread(file,',');
            delim=find(-1==line(:,1));
            data(n+1).mouse=line(1:delim(1)-1,:);
            data(n+1).paint=line((delim(1)+1):(delim(2)-1),:);
            data(n+1).mousedown=line((delim(2)+1):(delim(3)-1),:);
            data(n+1).mouseup=line((delim(3)+1):end,:);
        end
    end

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Stephen23 2023-1-11

编辑：Stephen23 2023-1-11

在 MATLAB Online 中打开

You might find that tidying up the code will help. A few tips:

** Always always always always always FCLOSE() every file that you FOPEN() !!!!

** Replace text concatenation of paths and filenames with FULLFILE(). For example, replace this:

[folder,'presentation.txt']

with

fullfile(folder,'presentation.txt')

** Align your code consistently, which helps when writing and debugging code. The MATLAB editor can do this for you: select all code, then press ctrl+i

** Note you can remove the dot-directories before the loop:

S = dir(..);
S(strncmp({S.name},'.',1)) = [];

** As the documentation recommends, consider replacing DLMREAD() with READMATRIX(), and TEXTSCAN() with READCELL() or READLINES() or whatever suits your data files.

** Preallocate arrays before loops: https://www.mathworks.com/help/matlab/matlab_prog/preallocating-arrays.html

Now lets start debugging: please show us the value of length(subjects)

请先登录，再进行评论。

请先登录，再回答此问题。

Follow Question