onnx-Ultra-Fast-Lane-Detection-Inference

Example scripts for the detection of lanes using the ultra fast lane detection model in ONNX.

Source: https://www.flickr.com/photos/32413914@N00/1475776461/

Pytorch inference

For performing the inference in Pytorch, check my other repository Ultrafast Lane Detection Inference Pytorch.

Tested Environment

Computer or Laptop

Single Board

• Jetson Xavier AGX. JetPack 4.6

Requirements for Laptop

• OpenCV, scipy, onnx and onnxruntime. pafy and youtube-dl are required for youtube video inference.

Requirements for Nvidia Xavier

• OpenCV, scipy, onnx and onnxruntime. pafy and youtube-dl, Nvidia Xavier AGX, JetPack 4.6 and Python3.6 .

Installation for Laptop

pip install -r requirements.txt
pip install pafy youtube-dl

Installation for Nvidia Xavier

pip3 install opencv-python
pip3 install scipy
Download >> https://nvidia.app.box.com/s/bfs688apyvor4eo8sf3y1oqtnarwafww
Install  >> pip3 install onnxruntime_gpu-1.8.0-cp36-cp36m-linux_aarch64.whl
pip3 install scikit-build
if you found not match version try to upgrade the PIP >> sudo -H pip3 install --upgrade pip

ONNX model

The original model was converted to different formats (including .onnx) by PINTO0309, download the models from his repository and save it into the models folder.

ONNX Conversion script: cfzd/Ultra-Fast-Lane-Detection#218

Original Pytorch model

The pretrained Pytorch model was taken from the original repository.

Model info (link)

• Input: RGB image of size 800 x 200 pixels.
• Output: Keypoints for a maximum of 4 lanes (left-most lane, left lane, right lane, and right-most lane).

Examples

• Image inference:

python imageLaneDetection.py 

• Webcam inference:

python webcamLaneDetection.py

• Video inference:

python videoLaneDetection.py

Inference video Example

Result for Nvidia Xavier

• YoutTube: https://www.youtube.com/watch?v=yBmcYDke7Wg
• Original video: https://youtu.be/2CIxM7x-Clc (by Yunfei Guo)