How to use Deepstream tracker when objects are similar in features and color

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU) Jetson NX

• DeepStream Version 6.0
• JetPack Version (valid for Jetson only) 4.6
• TensorRT Version 8.0.1
• NVIDIA GPU Driver Version (valid for GPU only)
• Issue Type( questions, new requirements, bugs) questions
• How to reproduce the issue ? (This is for bugs. Including which sample app is using, the configuration files content, the command line used and other details for reproducing)
• Requirement details( This is for new requirement. Including the module name-for which plugin or for which sample application, the function description)

I am currently using a tinyyolov4 model to perform object detection for person, hat, hivis, crane. See attached video. The issue I am having is when using the NvDCF tracker in Deepstream, the bounding box switches between the two crane objects when it moves from the top position to the bottom left shown in the following images. I have a feeling the crane objects are too similar in features and color and could be affecting the tracking ability. Also model is imperfect

image
image

I have attached my config files for the tracker.

tracker_config.yml (5.8 KB)
model_yolotiny.txt (757 Bytes)
dstest2_tracker_config.txt (551 Bytes)

Can I do anything else with the tracker config to improve this result?

3

Can you see the issue without tracker?

4

Sorry. I figured out my issue. My post-processing code was interfering with the output of the tracker hence why it was not working. I went back to the basics of running the deepstream-app in my PC using the nvcr.io/nvidia/deepstream:6.0-devel , and the tracker is working fine.

Thanks for replying.

6

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.