Skip to content

Latest commit

 

History

History
69 lines (38 loc) · 1.32 KB

README.md

File metadata and controls

69 lines (38 loc) · 1.32 KB
Raw

A baseline code for VSPW

Preparation

Download VSPW dataset

The VSPW dataset with extracted frames and masks is available here. Please download the 480p version of VSPW dataset.

Dependencies

  • Python 3.7
  • Pytorch 1.7
  • Numpy

Download the ImageNet-pretrained models at this link. Put it in the root folder and decompress it.

Train and Test

Edit the .sh files in scripts/ and change the $DATAROOT to your path to VSPW_480p.

Image-based methods

PSPNet

sh scripts/run_psp.sh

OCRNet

sh scripts/run_ocr.sh

Evaluation on TC and VC

Change dataroot and prediction root in TC_cal.py and VC_perclip.py.

python TC_cal.py

python VC_perclip.py

This implementation utilized this code and RAFT.

Citation

@inproceedings{miao2021vspw,

  title={VSPW: A Large-scale Dataset for Video Scene Parsing in the Wild},

  author={Miao, Jiaxu and Wei, Yunchao and  Wu, Yu and Liang, Chen and Li, Guangrui and Yang, Yi},

  booktitle={Proceedings of the {IEEE} Conference on Computer Vision and Pattern Recognition},

  year={2021}

}