Questions about get_converted_buffer in gstnvinfer.cpp

Accelerated Computing Intelligent Video Analytics DeepStream SDK
3
  1. For nvinfer plugin, the input video/image resolution can be different to network width and height, nvinfer plugin will do the pre-processing including scaling to meet the network requirment.
  2. Do you want to dump network input picture data? If so, please refer to DeepStream SDK FAQ - Intelligent Video Analytics / DeepStream SDK - NVIDIA Developer Forums, item 2.
  3. Do you mean when you by enabling “maintain-aspect-ratio”, you need the input image is centered? If so, current interfaces do not support it.
  4. The code for resizing and other preprocessing is in InferPreprocessor::transform() function in /opt/nvidia/deepstream/deepstream-5.0/sources/libs/nvdsinfer/nvdsinfer_context_impl.cpp. The algorithm is in /opt/nvidia/deepstream/deepstream-5.0/sources/libs/nvdsinfer/nvdsinfer_conversion.cu