This repository provides deployment examples for deep learning computer vision models using TensorRT. It implements CUDA-based image preprocessing, model inference, decode, and NMS operations optimized for GPU. The typical conversion pipeline is torch model to ONNX to TensorRT engine. Supported models include the YOLO series (YOLOv3 through YOLOv8), EfficientDet, RT-DETR, U2Net, and libfacedetection.