Trying to output coordinates from a CSV file and plot a CoG
Show older comments
Hi, I am trying to plot the CoG of 3 body landmarks stored in coordinates within a CSV file. However before I can plot the CoG, I need to import the CSV coordinates successfully for which I am struggling to acheive. Please see the attached matlab file and respond with any advice. Thanks. I am running Matlab on windows 11.
12 Comments
Mathieu NOE
on 3 Jan 2023
Edited: Mathieu NOE
on 3 Jan 2023
hi
it would help if you share the csv file along
have you tried with csvread ?
David Gill
on 3 Jan 2023
Edited: David Gill
on 3 Jan 2023
David Gill
on 3 Jan 2023
I cannot figure out what you are calculating, however using either readtable or readmatrix will be easier than what you’re currently doing —
T1 = readtable('https://www.mathworks.com/matlabcentral/answers/uploaded_files/1250332/object_cropDLC_resnet50_3_landmarks_for_whiskersDec13shuffle1_1030000.csv')
file = websave('plotting_CoG','https://www.mathworks.com/matlabcentral/answers/uploaded_files/1250297/plotting_CoG.m')
type(file)
M = table2array(T1(:,2:end))
ColumnMeans = mean(M)
Then do the other calculations on those values, if necessary. (I used readtable and then table2array here because I wanted to see what was in the table.)
.
Mathieu NOE
on 3 Jan 2023
your code is correct (if the intention was to have one CoG per landmark)
all the best
minor graphical output modifications :
% Read the data from the file into a matrix
data = csvread('object_cropDLC_resnet50_3_landmarks_for_whiskersDec13shuffle1_1030000.csv', 3, 0);
% Extract the x & y coordinates for the body landmarks from the matrix:
x1 = data(:,1);
y1 = data(:,2);
x2 = data(:,3);
y2 = data(:,4);
x3 = data(:,5);
y3 = data(:,6);
% Next, calculate the CoG of the body landmarks by averaging the x & y coordinates using the 'mean' function
CoGx = mean([x1 x2 x3]);
CoGy = mean([y1 y2 y3]);
% Plot the landmarks using the scatter function
figure;
scatter(x1, y1, 'r', 'o','filled');
hold on;
scatter(x2, y2, 'g', 'o','filled');
scatter(x3, y3, 'b', 'o','filled');
plot(CoGx, CoGy, '*k', 'Markersize',15);
legend({'Landmark 1', 'Landmark 2', 'Landmark 3', 'CoG'});
David Gill
on 3 Jan 2023
Image Analyst
on 3 Jan 2023
How about posting a screenshot of the data plotted, with an indication of where you think the centroid should approximately be?
David Gill
on 4 Jan 2023
Mathieu NOE
on 4 Jan 2023
simply change these lines
CoGx = mean([x1 x2 x3]);
CoGy = mean([y1 y2 y3]);
to
CoGx = mean([x1; x2; x3]);
CoGy = mean([y1; y2; y3]);
now you have 1 coG
David Gill
on 8 Jan 2023
Mathieu NOE
on 9 Jan 2023
hello
seems to me you have another post dedicated to this video processing issue (is not my expertise area I apologize)
maybe Image Analyst is again keen to help you on that topic
David Gill
on 9 Jan 2023
Accepted Answer
Image Analyst
on 9 Jan 2023
@David Gill try this:
% Demo to track and plot the weighted center of gravity of a video and create an output video.
% Demo by Image Analyst.
clc; % Clear the command window.
fprintf('Beginning to run %s.m ...\n', mfilename);
close all; % Close all figures (except those of imtool.)
clearvars;
workspace; % Make sure the workspace panel is showing.
format short g;
format compact;
fontSize = 14;
% Set the name and path of the input video file
inputVideoFileName = "G:\Whisker Tracking\DLC\3_landmarks_for_whiskers-David Gill-2022-12-13\object_cropDLC_resnet50_3_landmarks_for_whiskersDec13shuffle1_1030000_labeled.mp4";
% Set the name and path of the output video file
outputVideoFileName = 'G:\Whisker Tracking\DLC\outputCoG.mp4';
% Or, if you don't have your own video, use the sample video that ships with MATLAB.
% inputVideoFileName = 'rhinos.avi';
% outputVideoFileName = 'rhinos 2.avi';
% Read the input video file
inputVideo = VideoReader(inputVideoFileName)
% Create the output video file
outputVideo = VideoWriter(outputVideoFileName);
open(outputVideo);
% Define the region of interest to be the whole image.
mask = true(inputVideo.Height, inputVideo.Width);
% Extract the frames from the input video
hFig = figure;
frameCounter = 0;
while hasFrame(inputVideo)
thisFrame = readFrame(inputVideo);
frameCounter = frameCounter + 1;
imshow(thisFrame);
if size(thisFrame, 3) > 1
grayImage = rgb2gray(thisFrame);
else
grayImage = thisFrame;
end
% Get Weighted Centroid.
props = regionprops(mask, grayImage, 'WeightedCentroid');
xCentroid(frameCounter) = props.WeightedCentroid(1);
yCentroid(frameCounter) = props.WeightedCentroid(2);
% Plot the body landmarks and CoG on the frame
hold on;
scatter(xCentroid(frameCounter), yCentroid(frameCounter), 30, 'red', 'filled', 'o');
text(xCentroid(frameCounter), yCentroid(frameCounter), ...
' COG', 'VerticalAlignment', 'middle', 'Color', 'r', ...
'FontSize', fontSize, 'FontWeight','bold');
caption = sprintf('Frame #%d', frameCounter);
title(caption, 'FontSize', fontSize)
hold off;
drawnow; % Force immediate update
% Burn graphics into an output image. Requires the Computer Vision Toolbox.
outputFrame = insertMarker(thisFrame, [xCentroid(frameCounter), yCentroid(frameCounter)] ,'+','Color', 'r','size',30);
% Write the frame to the output video
writeVideo(outputVideo, outputFrame);
end
% Close the input video file
% close(inputVideo);
% Close the output video file
close(outputVideo);
% Close that figure and bring up another one where we can show the tracked centroid.
close(hFig);
hFig = figure;
subplot(2, 1, 1);
imshow(thisFrame);
hold on;
plot(xCentroid, yCentroid, 'r-', 'LineWidth', 2);
fontSize = 20
title('Path of Center of Gravity in Red', 'FontSize', fontSize)
% Plot centroid in next axes.
subplot(2, 1, 2);
plot(xCentroid, 'r-', 'LineWidth', 2);
hold on;
plot(yCentroid, 'g-', 'LineWidth', 2);
grid on;
legend('x centroid column', 'y centroid row', 'Location', 'west')
xlabel('Frame Number', 'FontSize', fontSize)
ylabel('Column (for x) or Row (for y)', 'FontSize', fontSize)
title('Weighted Center of Gravity (Weighted Centroid)', 'FontSize', fontSize)
% Maximize the figure.
hFig.WindowState = "maximized";
% Open the output video
if ispc
winopen(outputVideoFileName);
end
message = sprintf('Done processing %d frame.', frameCounter);
uiwait(helpdlg(message))
11 Comments
David Gill
on 9 Jan 2023
Edited: David Gill
on 11 Jan 2023
Image Analyst
on 11 Jan 2023
Then just use your precomputed centroids instead of computing new ones in the loop. In other words, delete these lines:
% Get Weighted Centroid.
props = regionprops(mask, grayImage, 'WeightedCentroid');
xCentroid(frameCounter) = props.WeightedCentroid(1);
yCentroid(frameCounter) = props.WeightedCentroid(2);
and just use your own
xCentroid(frameCounter) = CoGx(frameCounter);
yCentroid(frameCounter) = CoGy(frameCounter);
David Gill
on 13 Jan 2023
Image Analyst
on 13 Jan 2023
Please attach the video object_cropDLC_resnet50_3_landmarks_for_whiskersDec13shuffle1_1030000_labeled.mp4
If it's too big, zip it up. If still too big, tell me if I can use the standard rhinos.avi.
David Gill
on 13 Jan 2023
Image Analyst
on 13 Jan 2023
So do you just want to plot a spot on the image for (x1,y1), (x2, y2) and (x3, y3) for every frame in the image? Are you sure that the range of x and y are within your image? And are you sure there is an x and y for every frame?
David Gill
on 13 Jan 2023
Edited: David Gill
on 13 Jan 2023
Image Analyst
on 13 Jan 2023
Try this. I'm only plotting the average COG, not all 3 points.
% Demo by Image Analyst
clc; % Clear the command window.
close all; % Close all figures (except those of imtool.)
clear; % Erase all existing variables. Or clearvars if you want.
workspace; % Make sure the workspace panel is showing.
format short g;
format compact;
fontSize = 22;
markerSize = 15;
% Demo to track and plot the weighted center of gravity of a video and create an output video.
clc; % Clear the command window.
fprintf('Beginning to run %s.m ...\n', mfilename);
close all; % Close all figures (except those of imtool.)
clearvars;
workspace; % Make sure the workspace panel is showing.
format short g;
format compact;
fontSize = 14;
% Set the name and path of the input video file
folder = "G:\Whisker Tracking\DLC\3_landmarks_for_whiskers-David Gill-2022-12-13\";
inputVideoFileName = fullfile(folder, "\object_cropDLC_resnet50_3_landmarks_for_whiskersDec13shuffle1_1030000_labeled.mp4");
inputVideoFileName = 'rhinos.avi';
% Set the name and path of the output video file
outputVideoFileName = 'outputCoG.mp4'; % 'G:\Whisker Tracking\DLC\outputCoG.mp4';
% Read the input video file
inputVideo = VideoReader(inputVideoFileName)
% Create the output video file
outputVideo = VideoWriter(outputVideoFileName);
open(outputVideo);
% Read the data from the file into a matrix
data = csvread('object_cropDLC_resnet50_3_landmarks_for_whiskersDec13shuffle1_1030000.csv', 3, 0);
% Extract the x & y coordinates for the body landmarks from the matrix:
x1 = data(:,1);
y1 = data(:,2);
x2 = data(:,3);
y2 = data(:,4);
x3 = data(:,5);
y3 = data(:,6);
% Next, calculate the CoG of the body landmarks by averaging the x & y coordinates using the 'mean' function
CoGx = mean([x1, x2, x3], 2);
CoGy = mean([y1, y2, y3], 2);
%=================================================================================================
% SPECIAL CODE FOR RHINOS VIDEO ONLY!
% Scale so it fits in the image (Just for debugging using sample standard rhinos video ONLY!)
CoGx = CoGx(1:inputVideo.NumFrames);
CoGy = CoGy(1:inputVideo.NumFrames);
CoGx = rescale(CoGx, 1, inputVideo.Width);
CoGy = rescale(CoGy, 1, inputVideo.Height);
%=================================================================================================
% Plot the data
figure;
plot(x1, y1, '-', 'LineWidth', 2);
hold on;
plot(x2, y2, '-', 'LineWidth', 2);
plot(x3, y3, '-', 'LineWidth', 2);
legend('1', '2', '3')
grid on;
% Extract the frames from the input video
hFig = figure;
frameCounter = 1;
while hasFrame(inputVideo)
thisFrame = readFrame(inputVideo);
frameCounter = frameCounter + 1;
imshow(thisFrame);
if size(thisFrame, 3) > 1
grayImage = rgb2gray(thisFrame);
else
grayImage = thisFrame;
end
if frameCounter < numel(CoGy)
% Plot point if it's within the frame.
if CoGx(frameCounter) <= inputVideo.Width && CoGy(frameCounter) <= inputVideo.Height
% Plot the body landmarks and CoG on the frame
hold on;
scatter(CoGx(frameCounter), CoGy(frameCounter), 30, 'red', 'filled', 'o');
text(CoGx(frameCounter), CoGy(frameCounter), ...
' COG', 'VerticalAlignment', 'middle', 'Color', 'r', ...
'FontSize', fontSize, 'FontWeight','bold');
caption = sprintf('Frame #%d', frameCounter);
title(caption, 'FontSize', fontSize)
hold off;
drawnow; % Force immediate update
% Burn graphics into an output image. Requires the Computer Vision Toolbox.
outputFrame = insertMarker(thisFrame, [CoGx(frameCounter), CoGy(frameCounter)] ,'+','Color', 'r','size',30);
% Write the frame to the output video
writeVideo(outputVideo, outputFrame);
else
fprintf('Point (%.1f, %.1f) is outside the image, which is (%d, %d).\n', ...
CoGx(frameCounter), CoGy(frameCounter), inputVideo.Width, inputVideo.Height);
end
end
end
% Close the output video file
close(outputVideo);
% Maximize the figure.
hFig.WindowState = "maximized";
% Open the output video
outputVideoFileName = outputVideo.Filename
if ispc
winopen(outputVideoFileName);
end
close(outputVideo);
message = sprintf('Done processing %d frame.', frameCounter);
uiwait(helpdlg(message))
David Gill
on 14 Jan 2023
Image Analyst
on 14 Jan 2023
Then you better examine the x1, x2, and x3. The x1 looks especially suspicious.
David Gill
on 14 Jan 2023
More Answers (0)
Categories
Find more on Computer Vision with Simulink in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!
