Rescaling/padding while running inference in DS


I am using DS-5.1. I want to re-scale image and add padding to maintain the aspect ratio. I am using people-net model for object detection model and one classification which has 224*224 size and has trained with padding (To maintain aspect ratio). Now I want to use that model and want to re-scale/padding to image before classification and after detection (person b-box). How we can set in configuration parameter or we need to add plugin for that in between detection and classification ?


We have peoplenet sample in NVIDIA-AI-IOT/deepstream_tao_apps at release/tlt3.0 (