How should I understand the different set for depth(3) and pose(5) in --seq_length?

yzcjtr / GeoNet

Code for GeoNet: Unsupervised Learning of Dense Depth, Optical Flow and Camera Pose (CVPR 2018)

MIT License

726 stars 181 forks source link

How should I understand the different set for depth(3) and pose(5) in --seq_length? #10

Closed SlingHe closed 6 years ago

SlingHe commented 6 years ago

In the tutorial, you mention that

Note that for replicating depth and pose results, the --seq_length is suggested to be 3 and 5 respectively.

However in your source code, depth and pose estimation are trained together. How can you set different seq_length for the whole pipeline? I'm confused for a while.

kristijanbartol commented 6 years ago

I guess the best depth results are obtained using --seq_length=3 and separately, the best pose results are obtained using --seq_length=5. I don't think you can have both in the same pipeline.

yzcjtr commented 6 years ago

@kristijanbartol is right. They are trained with different sequence lengths as well as splits. You could read the experimentation part of our paper for more details.