• State
  • Locked Locked
  • Replies 2 replies
  • Subscribers 100 subscribers
  • Views 124 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Original question:

SK-AM62A-LP: Model Compilation and Deployment using Model Composer

SK-AM62A-LP: Custom Model Inferencing step

Anusha Subramanyam
Prodigy 80 points

Part Number: SK-AM62A-LP

Tool/software:

Hi,

I have trained and compiled the object detection model using Model Composer SDK 9.1. The Live Preview is functioning as expected.

The model is deployed on the AM62A starter kit and executed with the following command:

         ./app_edgeai.py ../configs/my_configuration.yaml

I have attached the screenshot of my_configuration.yaml file along with the log file for reference.

When setting kmssink to 640x480, the pipeline initializes but then de-initializes with the error: [ERROR] GStreamer error: negotiation problem.

Observations:

  1. When  input stream resolution is set to 640x640 because the model is trained on the yolox_s_lite model. If a different resolution is used, the following error occurs: [ERROR] Could not get allowed GstCaps of device.
  2. If kmssink output resolution is set to 1980x1080, the following error is encountered: [ERROR] Flow output resolution cannot be greater than input resolution.

Could you please help me understand the cause of these issues and suggest possible resolutions?

Regards,

Anusha