How to set true center crop for classification model in deepstream pipeline?

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

• Hardware Platform (Jetson / GPU):jetson
• DeepStream Version:5.1
• JetPack Version (valid for Jetson only):4.5
• TensorRT Version:7.1.3

Hi,
I trained resnet-18 classification with TLT3.0 on my own custom dataset, and in train and evaluation phases center crop is True and applied, If I set center crop to False in the train and evaluation phases, the accuracy is drop incredibly, So I have to set True for center crop in train and evaluation and inference phases. I want to use the trained model in the deepstream pipeline.
My question is that How I can use center crop in the nvinfer property for this model and apply center crop in the pipeline.
Is it possible? How?

3

Current deepstream does not support such preprocessing.

5

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.