/Center-and-Scale-Prediction-CSP-Pytorch

Pytorch implementation of CSP

Primary LanguagePython

Contents

Installation

  1. Install Pytorch>=1.0.0 in python3.
  2. Clone this repository.
git clone https://github.com/zhangminwen/Center-and-Scale-Prediction-CSP-Pytorch.git

Testing

  1. Download our pre-trained model and save in $CSP_ROOT/weights. model:epoch320 password:s06j
  2. Test the model by using gpu.
cd $CSP_ROOT
python ./test.py
  1. Test the model by using cpu.
python ./test.py --cpu

The "result.jpg" is detection result.

Training

  1. Prepare the dataset in $CSP_ROOT/city.

    The pedestrian annotations (xmin, ymin, xmax, ymax) in each image are stored in the txt document which name correspond to the image name and saved in $CSP_ROOT/city/Annotations.

    Images are saved in $CSP_ROOT/city/Images.

    Splits of training, testing, valudation are saved in $CSP_ROOT/city/ImageSets.

  2. Download pretrained ResNet50 model and save in $CSP_ROOT/model.

  3. Train

  python ./train.py

References

@inproceedings{liu2018high,
  title={High-level Semantic Feature Detection: A New Perspective for Pedestrian Detection},
  author={Wei Liu, Shengcai Liao, Weiqiang Ren, Weidong Hu, Yinan Yu},
  booktitle={IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year={2019}
}