Parallel Processing and Preallocating for Speed

Question

Jacob Mevorach 2017-3-30

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/332782-parallel-processing-and-preallocating-for-speed

评论： Jacob Mevorach 2017-4-17

So I've been trying to implement parfor in the following for loop

  function [tracks_Out, nextId_Out] = createNewTracks(tracks, centroids, bboxes, unassignedDetections, nextId)
      centroids = centroids(unassignedDetections, :);
      bboxes = bboxes(unassignedDetections, :);
      tracks_Out=tracks;
      nextId_Out=nextId;
        for i = 1:size(centroids, 1)
            centroid = centroids(i,:);
            bbox = bboxes(i, :);
            % Create a Kalman filter object.
            kalmanFilter = configureKalmanFilter('ConstantVelocity', ...
                centroid, [200, 50], [100, 25], 100);
            % Create a new track.
            newTrack = struct(...
                'id', nextId_Out, ...
                'bbox', bbox, ...
                'kalmanFilter', kalmanFilter, ...
                'age', 1, ...
                'totalVisibleCount', 1, ...
                'consecutiveInvisibleCount', 0);
            % Add it to the array of tracks.
            tracks_Out(end + 1) = newTrack;
            % Increment the next id.
            nextId_Out = nextId_Out + 1;
        end
    end

The line with tracks_Out(end + 1) gives me the message the variable "tracks_Out appears to change size on every loop iteration, consider preallocating for speed". I think I get that its saying I should have some sort of statement where I say tracks out = zeros(1,1000000) or something to that affect where I pad tracks_Out with a bunch of zeros at the end of it every time but for some reason I'm having trouble understanding how I could do that without leaving a bunch of unnecessary zeros in my tracks variable.

Also, I can't use parfor on this loop because tracks_Out isn't being indexed with the loop variable. I have a feeling that both the parfor problem and the preallocating problem could be solved in the same go with the same solution but I'm having trouble seeing it right now. If anyone could see a way I could do that I'd greatly appreciate it.