Skip to content

Conversation

@khan-yin
Copy link
Contributor

@khan-yin khan-yin commented May 22, 2023

Environment

  • PaddlePaddle2.4.0
  • V100
  • need pip install imgaug>=0.4.0 for image augmentations in requirements.txt
imgaug>=0.4.0

Training

  • single GPU
python tools/train.py -c configs/clrnet/clr_resnet18_culane.yml -o use_gpu=true
  • multi GPU
export CUDA_VISIBLE_DEVICES=0,1,2,3
python -m paddle.distributed.launch --gpus 0,1,2,3 tools/train.py -c configs/clrnet/clr_resnet18_culane.yml

P.S. recommend single GPU training for better accuracy.

Evaluation

mkdir evaluation
python tools/eval.py -c configs/clrnet/clr_resnet18_culane.yml -o weights=output/clr_resnet18_culane/model_final.pdparams output_eval=evaluation/ use_gpu=true

Inference

python tools/infer_culane.py -c configs/clrnet/clr_resnet18_culane.yml -o weights=output/clr_resnet18_culane/model_final.pdparams use_gpu=true --infer_img=demo/lane00000.jpg

Weights&Logs

backbone mF1 F1@50 F1@75 download code
ResNet-18 54.98 79.46 62.10 BaiduDisk fz71
.
|-- model_final.pdparams
|-- evaluation.log
|-- training.log

Demo

lane00000_demo

@paddle-bot
Copy link

paddle-bot bot commented May 22, 2023

Thanks for your contribution!

@khan-yin khan-yin requested a review from LokeZhou May 30, 2023 06:47
@jerrywgz jerrywgz merged commit 129ddbb into PaddlePaddle:develop Jun 1, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants