Tao Segformer inference on tensorRT preprocessing and post processing

on tao 5.2, segformer, model mit_b5 with rgb images

Looking for documentation on

a) Preprocessing the video frame to normalize and NHWC or NCHW

b) Post-processing the inference buffer to get segmentation values for each pixel.


You can refer to tao_deploy/nvidia_tao_deploy/cv/segformer/inferencer.py at main · NVIDIA/tao_deploy · GitHub.