Does the secondary GIE resizes its input to match classifier input dimension?

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU): Jetson AGX Xavier
• DeepStream Version: 5.0
• JetPack Version (valid for Jetson only): 4.4
• TensorRT Version: 10.2
• Issue Type( questions, new requirements, bugs): question

I have classifier trained via TLT, before training, all images are resized to match the classifier input size so I wonder in deepstream, beside cropping (according to the metadata from primary detector), does the secondary GIE does any resizing internally before perform classification?

3

Yes, nvinfer will do the resize based on the infer-dims you specified in nvinfer config file , pls check Gst-nvinfer — DeepStream 6.1.1 Release documentation for more details