How to write helper functions in Lidar 3-D object detections

Can anyone help me with the getting these helper functions in matlab, cropFrontViewFromLidarData helper function, saveptCldToPCD helper function, generateGTDataForAugmentation helper function, groundTruthDataAugmentation helper function,

[Pavan Kumar B N]

Hi @Ganesh Sai Doggali are you able to detect objects in LiDAR data? Could you please provide some details about your implementation? I am also working on the same. Thank you.

[Shahana Bano]

function [croppedPointCloudObj, processedLabels] = cropFrontViewFromLidarData(lidarData, boxLabels, gridParams)

% This function crops the front view from the input full-view point cloud

% and also processes the corresponding box labels according to the

% specified grid parameters.

tmpStr = '';

numFiles = size(boxLabels,1);

processedLabels = cell(size(boxLabels));

croppedPointCloudObj = cell(size(numFiles));

% Get the classes from the ground truth labels.

classNames = boxLabels.Properties.VariableNames;

for i = 1:numFiles

ptCloud = read(lidarData);

groundTruth = boxLabels(i,:);

% Set the limits for the point cloud.

[x,y] = find( ptCloud.Location(:,:,1) < gridParams{1,2}{1} ...

& ptCloud.Location(:,:,1) > gridParams{1,1}{1} ...

& ptCloud.Location(:,:,2) < gridParams{1,2}{2} ...

& ptCloud.Location(:,:,2) > gridParams{1,1}{2} ...

& ptCloud.Location(:,:,3) < gridParams{1,2}{3} ...

& ptCloud.Location(:,:,3) > gridParams{1,1}{3});

ptCloud = select(ptCloud, x, y, 'OutputSize', 'full');

processedData = removeInvalidPoints(ptCloud);

for ii = 1:numel(classNames)

labels = groundTruth(1,classNames{ii}).Variables;

if(iscell(labels))

labels = labels{1};

end

if ~isempty(labels)

% Get the label indices that are in the selected RoI.

labelsIndices = labels(:,1) > gridParams{1,1}{1} ...

& labels(:,1) < gridParams{1,2}{1} ...

& labels(:,2) > gridParams{1,1}{2} ...

& labels(:,2) < gridParams{1,2}{2} ...

& labels(:,4) > 0 ...

& labels(:,5) > 0 ...

& labels(:,6) > 0;

labels = labels(labelsIndices,:);

if ~isempty(labels)

% Find the number of points inside each ground truth

% label.

numPoints = arrayfun(@(x)(findPointsInsideCuboid(cuboidModel(labels(x,:)),processedData)),...

(1:size(labels,1)).','UniformOutput',false);

posLabels = cellfun(@(x)(length(x) > 50), numPoints);

labels = labels(posLabels,:);

end

end

processedLabels{i,ii} = labels;

end

croppedPointCloudObj{i,1} = processedData;

end

% Print completion message when done.

msg = sprintf('Processing data 100%% complete');

fprintf(1,'%s',[tmpStr, msg]);

processedLabels = cell2table(processedLabels);

numClasses = size(processedLabels,2);

for j = 1:numClasses

processedLabels.Properties.VariableNames{j} = classNames{j};

end

end

[Ramya Sivanandham]

Thank you @Shahana Bano. I have tried this function. But im getting some parse error. If possible, could you help me to solve this issue?

rror using matlab.io.datastore.FileDatastore/read (line 29)

Error using ReadFcn @(x)pcread(x) for file:

C:\ .....\Lidar_Dataset\Pandaset\Lidar\0472.pcd

Error using visionPcdRead

Internal file parsing error. Please see the documentation for more details.

Error in pcread (line 99)

properties = visionPcdRead(filename,requiredProperties,optionalProperties);

Error in main_Lidar>@(x)pcread(x) (line 33)

lidarData = fileDatastore(path,'ReadFcn',@(x) pcread(x));

Error in cropFrontViewFromLidarData (line 17)

ptCloud = read(lidarData);

Error in main_Lidar (line 83)

[croppedPointCloudObj,processedLabels] = cropFrontViewFromLidarData(lidarData,boxLabels,gridParams);

29 [data, info] = matlab.io.datastore.FileBasedDatastore.errorHandlerRoutine(fds,e);

473 throwAsCaller(ME);

K>>

[Ramya Sivanandham]

Though, that particular file is available in that path, this error still exists...

Answer 1

[Mahesh Taparia]

Hi

The helper functions are shipped with the example. You can check the helper functions in the example folder by using command mentioned in that particular example. In your case, you are referring to this pointPillars example, whose command is present at the right side of the documentation (as View MATLAB Command), which is given below:

openExample('deeplearning_shared/Lidar3DObjectDetectionUsingPointPillarsExample')

Run this command, it will redirect to respective folder where all the associated functions are present.

Hope it will help!

[Ganesh Sai Doggali]

Thank You !