PWC Net
1.0.0
版權(C)2018 NVIDIA Corporation。版權所有。根據CC BY-NC-SA 4.0許可(https://creativecommons.org/licenses/by-nc-sa/4.0/legalcode)獲得許可。
對於Caffe用戶,請參閱Caffe/readme.md。
有關Pytorch用戶,請參考pytorch/readme.md
Pytorch的實施幾乎與CAFFE實施相匹配(Sintel訓練集的最終傳球:Pytorch的2.31,Caffe的平均EPE)。
PWC-NET融合了幾種經典的光流估計技術,包括圖像金字塔,翹曲和成本量,以端到端可訓練的深度神經網絡,以實現最先進的結果。

Deqing Sun,Xiaodong Yang,Ming-Yu Liu和Jan Kautz。 “ PWC-NET:使用金字塔,翹曲和成本量的光流的CNN。” CVPR 2018或ARXIV:1709.02371
更新和擴展版本:“模型很重要,培訓也是如此:CNN的經驗研究,用於光流估計。” Arxiv:1809.05571
項目頁面鏈接
在強大的視力挑戰研討會上談論
在CVPR 2018會議上談話
如果您使用PWC-NET,請引用以下論文:
@InProceedings{Sun2018PWC-Net,
author = {Deqing Sun and Xiaodong Yang and Ming-Yu Liu and Jan Kautz},
title = {{PWC-Net}: {CNNs} for Optical Flow Using Pyramid, Warping, and Cost Volume},
booktitle = CVPR,
year = {2018},
}
或arxiv紙
@article{sun2017pwc,
author={Sun, Deqing and Yang, Xiaodong and Liu, Ming-Yu and Kautz, Jan},
title={{PWC-Net}: {CNNs} for Optical Flow Using Pyramid, Warping, and Cost Volume},
journal={arXiv preprint arXiv:1709.02371},
year={2017}
}
或更新和擴展版本
@article{Sun2018:Model:Training:Flow,
author={Sun, Deqing and Yang, Xiaodong and Liu, Ming-Yu and Kautz, Jan},
title={Models Matter, So Does Training: An Empirical Study of CNNs for Optical Flow Estimation},
journal={IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)},
note = {to appear}
}
對於多幀流程,也請引用
@inproceedings{ren2018fusion,
title={A Fusion Approach for Multi-Frame Optical Flow Estimation},
author={Ren, Zhile and Gallo, Orazio and Sun, Deqing and Yang, Ming-Hsuan and Sudderth, Erik B and Kautz, Jan},
booktitle={Proceedings of the IEEE Winter Conference on Applications of Computer Vision (WACV)},
year={2019}
}
Fownet2-Pytorch
通過移動攝像頭進行3D運動場估計的動態場景中的剛性(ECCV 2018)
deqing sun([email protected])