Skip to content

v1.0.0

Latest
Latest

Choose a tag to compare

View all tags
@HenrikTrom HenrikTrom released this 27 May 15:45
· 18 commits to main since this release
v1.0.0
ca180a6

🐳 TRT-OpenCV-Dev v1.0.0

DOI

A Docker development environment for building high-performance C++ modules with TensorRT and OpenCV.

Customizable Docker container with support for CUDA, cuDNN, TensorRT, OpenCV, and CMake — tailored for efficient deployment and development on NVIDIA GPUs.

📑 Citation

If you use this software, please use the GitHub “Cite this repository” button at the top(-right) of this page.

Installation

Clone the repository and initialize submodules:

git clone --recurse-submodules https://github.com/HenrikTrom/Docker-OpenCV-TensorRT-Dev
git submodule update --init --remote --recursive

1: Configure .env Parameters

Edit the .env file in the root directory and set the following variables:

UBUNTU_IMAGE_VERSION=22.04
CUDA_VERSION=12.3
TENSORRT_VERSION=8.6.1.6
CUDA_ARCH_BIN=8.6
OPENCV_VERSION=4.10.0
UID=1000         # id -u
GID=1000         # id -g
CMAKE_VERSION=3.28.0  # Must be >= 3.27.7
CPP_OPTIMIZATIONS="-O3 -march=native"

Important:
Ensure no conflicting environment variables are defined in your shell.
Docker will override variables in your .env file if environment values exist.

2: Get TensorRT

  1. Download the matching TensorRT .tar.gz archive.

  2. Place it in:

    ./build/vision_dependencies/tensorrt/

  3. Verify that the file name matches the expected value in:

    ./build/vision_dependencies/tensorrt/install.sh

🧪 Tested with

  • Ubuntu 20.04, CUDA 11.8, TensorRT 8.6.1.6, OpenCV 4.10.0
  • Docker Engine 24+
Assets 2
Loading