A PyTorch reimplementation of the Omnidepth paper from Zioulis et al., ECCV 2018:
Notable difference with the paper: PyTorch's weight decay for the Adam solver does not seem to function the same way as Caffe's. Hence, I do not use weight decay in training. Instead, I use a learning rate schedule
this project is riched by some trainning visualizations based on others
- Python 3.7
- Pytorch 1.1
- Visdom 0.1.8
-
If you do use this repository, please make sure to cite the authors' original paper: Zioulis, Nikolaos, et al. "OmniDepth: Dense Depth Estimation for Indoors Spherical Panoramas." Proceedings of the European Conference on Computer Vision (ECCV). 2018.