Object detection and tracking using lidar point cloud data

Hello All,

I am working on object detection and tracking using veloyne lidar point cloud data (PCAP) without using Deep learning. I am a beginner, please guide me how to start with.

Thanks in advance

Answer 1

[Manish]

Hi Pavan,

I understand that you want to detect and track the lidar point clouds without using deep learning, you can follow these steps:

1. Preprocess Your Point Cloud: Begin by preprocessing your point cloud data according to your specific use case. You can use functions such as ‘pcorganize’ or ‘pcfitplane’ to organize the data or fit planes, respectively.
2. Segment the Point Cloud: Utilize the ‘pcsegdist’ function to segment the point cloud into clusters. This function groups the points based on Euclidean distance.
3. Extract Desired Clusters: Once the segmentation is complete, extract the clusters that are relevant to your application from the output of ‘pcsegdist’.
4. Object Detection: Pass the extracted clusters to the ‘objectDetection’ constructor.
5. Track Objects: Use the ‘trackerJPDA’ (Joint Probabilistic Data Association) to track the detected objects.

Here is the example which follows the above workflow:

• https://www.mathworks.com/help/vision/ug/track-vehicles-using-lidar.html

Refer to the Links below for better understanding:

• https://www.mathworks.com/help/vision/ref/pcsegdist.html
• https://www.mathworks.com/help/fusion/ref/objectdetection.html
• https://www.mathworks.com/help/fusion/ref/trackerjpda-system-object.html

Hope this helps!

Answer 2

[Joseph]

Object tracking using Veloyne lidar point cloud data requires a step-by-step workflow and uses functions like ' pcorganize ' or ' pcfitplane ' to align the data or fit to planes respectively.

following this example: https://www.mathworks.com/help/vision/ug/track-vehicles-using-lidar.html/happy wheels