Home

ViP the deep learning platform for all video-based networks.

Why?

It provides an easy to use and efficient pipeline to load data, train models, and evaluate results. Additionally it has an ever growing selection of video network architectures for action recognition and video object detection.

Table of Contents

• Common Use Cases
  • Test existing models on a dataset
  • Train existing models on a dataset
  • Add a model
• Preprocessing Options
• Add a Dataset

ViP Workflow

Citation

Please cite ViP when releasing any work that used this platform: https://arxiv.org/abs/1910.02793

@article{ganesh2019vip,
  title={ViP: Video Platform for PyTorch},
  author={Ganesh, Madan Ravi and Hofesmann, Eric and Louis, Nathan and Corso, Jason},
  journal={arXiv preprint arXiv:1910.02793},
  year={2019}
}

Acknowledgements

• C3D - https://github.com/jfzhang95/pytorch-video-recognition/
• I3D - https://github.com/piergiaj/pytorch-i3d
• SSD - https://github.com/amdegroot/ssd.pytorch
• DVSA (+fw, obj) - https://github.com/MichiganCOG/Video-Grounding-from-Text