• State
  • Locked Locked
  • Replies 1 reply
  • Subscribers 100 subscribers
  • Views 126 views
  • Users 0 members are here
Support feedback
Options
Options

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.

TDA4VM: There is a significant difference in the YOLOv5s model results between ARM mode and TIDLNode when compiled using edgeai tidl tools.

xiu yin
Prodigy 50 points
Part Number: TDA4VM

Tool/software:

Hi,

I trained a YOLOv5s model. In order to improve accuracy, I removed the post-processing part when compiling the model. When using edgeai-tidl-tools, I first run python3 onnxrt_ep.py - d and output the model results in arm mode. Run python3 onnxrt_ep. py - c again and output the model results using tidlnode. There is a significant difference in the model results output by the two methods, with - d being the correct result and - c only partially correct.
Later, I set the debug level to 4 and tensororbits to 32 in common_util.py, but the model output was completely incorrect.
May I ask where the problem may be? How to be a bug?

Regards,

Yolanda