search

SenZHANG-GitHub / ekf-imu-depth

[ECCV 2022] Towards Scale-Aware, Robust, and Generalizable Unsupervised Monocular Depth Estimation by Integrating IMU Motion Dynamics
194 stars 13 forks source link

Pretrained Models #2

Closed ldkong1205 closed 1 year ago

ldkong1205 commented 1 year ago

Hi @chaimi2013 @SenZHANG-GitHub, thank you for open-sourcing your work!

Would you like to release your pretrained weights for reproduction? Thanks!

SenZHANG-GitHub commented 1 year ago

Hi @chaimi2013 @SenZHANG-GitHub, thank you for open-sourcing your work!

Would you like to release your pretrained weights for reproduction? Thanks!

Sure. We will release the weights shortly in this month.

ldkong1205 commented 1 year ago

Thanks a lot and looking forward to it!

SenZHANG-GitHub commented 1 year ago

Hi @chaimi2013 @SenZHANG-GitHub, thank you for open-sourcing your work!

Would you like to release your pretrained weights for reproduction? Thanks!

Hi @ldkong1205 , thanks for your reminder and sorry for the delay! I have updated README.md to include detailed data preprocessing steps and the links to the pretrained models (DynaDepth R18 and R50 that use the ResNet18 and ResNet50 as the backbone, respectively).

SenZHANG-GitHub commented 1 year ago

Hi @chaimi2013 @SenZHANG-GitHub, thank you for open-sourcing your work!

Would you like to release your pretrained weights for reproduction? Thanks!

As for the preprocessing part, since the unsync kitti dataset is quite large. We also have updated the link to our preprocessed imu files in README.md. Feel free to check them : )