jetson-nano-dev

Docker build environment based on ubuntu 20.04 meant for computer vision / ML applications on Nvidia's jetson nano board, aside with an x86 (CPU only) test and run evirnonment for local testing.

Content

The following packages are installed:

• CMake 3.23
• OpenCV 4.7 with GPU support
• Librealsense 2.53 with GPU support (from repo)
• Cuda toolkit 10.2
• TensorRT: 8.2
• Tiny tensor: TensorRT wrapper
• OnnxRuntime: 1.11 with TensorRT support

Pre-requisites

sudo apt-get install qemu binfmt-support qemu-user-static

Variants

• jetson-nano-dev-arm-20-build: default build environment
• jetson-nano-dev-x86-20-build: x86 test environment with only CPU support to test algorithms locally
• jetson-nano-dev-x86-20-run: runtime env based on the build container

How to build

make build-all-20

Docker-hub

• https://hub.docker.com/r/attiladoor/jetson-nano-dev-arm-20-build
• https://hub.docker.com/r/attiladoor/jetson-nano-dev-x86-20-build
• https://hub.docker.com/r/attiladoor/jetson-nano-dev-x86-20-run

See the corresponding release tags