Assistance Required for ROI Configuration with nvpreprocess in DeepStream

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Hi

I hope this message finds you well.

I am currently working on a DeepStream 7.1 pipeline where I want a specific model to run only within a defined Region of Interest (ROI). I have used the nvpreprocess plugin for this purpose, but it seems the configuration is not functioning as expected. The model still processes the entire frame instead of being limited to the ROI.

Below is the configuration I used for both nvpreprocess and the PGIE. I would appreciate it if you could review the settings and let me know if there is anything missing or incorrectly set.

config_detect_face.txt (875 Bytes)
config_preprocess.txt (1.6 KB)

Thank you in advance for your support.

3

Please provide complete information as applicable to your setup.
• Hardware Platform (Jetson / GPU)
• DeepStream Version
• JetPack Version (valid for Jetson only)
• TensorRT Version
• NVIDIA GPU Driver Version (valid for GPU only)
• Issue Type( questions, new requirements, bugs)
• 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)

4

DeepStream 7.1
Hardware Platform GPU

5
  1. Please refer to the ready-made sample /opt/nvidia/deepstream/deepstream/sources/apps/sample_apps/deepstream-preprocess-test.
  2. can you see the output bboxes? If the issue still persists, could you share a complete running log?
6

The pipeline functions correctly without any problems, but the inference is performed on the entire frame instead of only within the ROI.

7 
    self.streammux.link(queue0)
    queue0.link(self.nvvideoconvert)
    self.nvvideoconvert.link(filter1)
    filter1.link(self.pgie)
    self.pgie.link(self.tracker)
    self.tracker.link(nvanalytics)
    nvanalytics.link(queue1)
    queue1.link(self.preprocess)
    self.preprocess.link(self.pgie1)
    self.pgie1.link(self.sgie2)
    self.sgie2.link(self.sgie4)
8
  1. are there any bboxes outside the ROI? could you share a complete running log? could you dump the media pipeline by referring to this faq.
  2. can you use deepstream-preprocess-test to reproduce this issue? if not, please compare the cfgs with yours. if so, please provide the cfg files.
9

Please refer to this topic. please set input-tensor-from-meta=1 in nvinfer cfg file.

10

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