YOLO11n engine running on Jetson nano
This video is about running YOLO11n inference on Jetson nano optimized for tensorrt.
All image processing, detection is running on the edge device and only BB labeled images are published via websocket.
Entire video is in real time.
$ jetson_release
Software part of jetson-stats 4.3.1 - (c) 2024, Raffaello Bonghi
Model: NVIDIA Jetson Nano Developer Kit - Jetpack 4.6.6 [L4T 32.7.6]
NV Power Mode[0]: MAXN
Serial Number: [XXX Show with: jetson_release -s XXX]
Hardware:
P-Number: p3448-0002
Module: NVIDIA Jetson Nano module (16Gb eMMC)
Platform:
Distribution: Ubuntu 18.04 Bionic Beaver
Release: 4.9.337-tegra
jtop:
Version: 4.3.1
Service: Active
Libraries:
CUDA: 10.2.300
cuDNN: 8.2.1.32
TensorRT: 8.2.1.9
VPI: 1.2.3
Vulkan: 1.2.70
OpenCV: 3.4.6 - with CUDA: YES
Steps to follow
1. Install Python 3.8 and make venv
2. With venv on, install pytorch along with torchvision 0.9.0 (Download below wheel built from the source for aarch64)
https://drive.google.com/file/d/16Jd7...
3. With venv on, install ultralytics from github
4. Install tensorrt python binding using below wheel file.
https://drive.google.com/file/d/1v0R4...
5. Now ultralytics, YOLO can be imported in python 3.8
For details follow instruction below.
https://www.notion.so/Install-ultraly...