How can I apply my faster RCNN Object Detector on the example "Tracking Pedestrians from a Moving Car"

1 回表示 (過去 30 日間)
I trained my faster RCNN Object Detector. I want to apply my detector on a video. So, I found a good example in Matlab. It's the Tracking Pedestrians from a Moving Car Example. I tried to change the ACF detector to the faster RCNN detector in the example. But It couldn't work. Please let me know how to apply the faster RCNN detector on the example. Thank you!
% Create system objects used for reading video, loading prerequisite data file, detecting pedestrians, and displaying the results.
videoFile = 'vippedtracking.mp4';
scaleDataFile = 'pedScaleTable.mat'; % An auxiliary file that helps to determine the size of a pedestrian at different pixel locations.
obj = setupSystemObjects(videoFile, scaleDataFile);
detector = peopleDetectorACF('caltech');
  2 件のコメント
Birju Patel
Birju Patel 2017 年 7 月 20 日
When you say it's not working, do you mean it doesn't detect anything or are you getting an error?
Jungmoon Kim
Jungmoon Kim 2017 年 8 月 1 日
Thank you for your comment.
The modifications in the code are shown below. I modified the variable named "detector".
The first picture is the pre-modification variable,
and the second picture is the post-modification variable.
What I want is to detect the video with faster R-CNN.
When I posted this question, I got an error in my code. I checked the code again today and made some changes.
if true
[bboxes, scores] = detect(detector, frame, option.ROI); %, ...
'WindowStride', 2,...
'NumScaleLevels', 4, ...
'SelectStrongest', false);
end
In this code we have removed "WindowStride, NumScaleLevels" part because we got an error because of "WindowStride, NumScaleLevels".
When I run the code again, it will just run. But there are some problems. Execution speed is too slow. And detection is not working.
What I have to do?

サインインしてコメントする。

回答 (0 件)

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!