Deep Learning API and Server in C++14 support for Caffe, PyTorch,TensorRT, Dlib, NCNN, Tensorflow, XGBoost and TSNE

FastFlowNet: A Lightweight Network for Fast Optical Flow Estimation (ICRA 2021)

BEVDet implemented by TensorRT, C++； Achieving real-time performance on Orin

Deploy stable diffusion model with onnx/tenorrt + tritonserver

Hardware-accelerated DNN model inference ROS 2 packages using NVIDIA Triton/TensorRT for both Jetson and x86_64 with CUDA-capable GPU

Yolov5 TensorRT Implementations

ComfyUI Depth Anything Tensorrt Custom Node (up to 5x faster)

Using TensorRT for Inference Model Deployment.

you can use dbnet to detect word or bar code,Knowledge Distillation is provided,also python tensorrt inference is provided.

this is a tensorrt version unet, inspired by tensorrtx

Production-ready YOLO8 Segmentation deployment with TensorRT and ONNX support for CPU/GPU, including AI model integration guidance for Unitlab Annotate.

VitPose without MMCV dependencies

3d object detection model smoke c++ inference code

C++ TensorRT Implementation of NanoSAM

The real-time Instance Segmentation Algorithm SparseInst running on TensoRT and ONNX

Base on tensorrt version 8.2.4, compare inference speed for different tensorrt api.

Convert yolo models to ONNX, TensorRT add NMSBatched.

DepthStream Accelerator: A TensorRT-optimized monocular depth estimation tool with ROS2 integration for C++. It offers high-speed, accurate depth perception, perfect for real-time applications in robotics, autonomous vehicles, and interactive 3D environments.