A repository for developing the YOLOPv2 algorithm in real time using a webcam.
Adapted from https://github.com/CAIC-AD/YOLOPv2
This code is developed to detect lane & drivable area in Autonomous driving competition. Made by Byounggun Park(Comflife).
(check your cuda, pytorch version before install)
conda create -n realtime python=3.8.16
conda activate realtime
conda install pytorch torchvision torchaudio pytorch-cuda=11.8 -c pytorch -c nvidia
conda install matplotlib
conda install scikit-learn
pip install opencv-python
(check your camera number --source 0,1,2)
`cd Realtime_yolopv2/ && python realtime.py --source 0`
You can get the model from here. This is the original YOLOPv2 model and this was trained by BDD100k dataset.
YOLOPv2: Better, Faster, Stronger for Panoptic Driving Perception