MarSaKi / VLN-BEVBert

[ICCV 2023} Official repo of "BEVBert: Multimodal Map Pre-training for Language-guided Navigation"
185 stars 5 forks source link

img_features/mp3d_224x224_vfov90_bgr.hdf5 files #2

Closed statjuns closed 1 year ago

statjuns commented 1 year ago

Thank you for sharing your nice work!!

I'm trying to run the training code but I cannot find img_features/mp3d_224x224_vfov90_bgr.hdf5 and img_features/habitat_224x224_vfov90_depth.hdf5. How can I get the files?

MarSaKi commented 1 year ago

Thanks for your interest. These two files are for visualization not for training. You can comment corresponding vis code if you met errors during training. If you want to visualize bev maps, you can refer to the “grid_mp3d_clip.py” and “grid_depth.py” files to directly store RGB and depth images.

MarSaKi commented 1 year ago

Thank you for sharing your nice work!!

I'm trying to run the training code but I cannot find img_features/mp3d_224x224_vfov90_bgr.hdf5 and img_features/habitat_224x224_vfov90_depth.hdf5. How can I get the files?

I have uploaded image storing scripts in "precompute_features" folder, check it for your convenience.

statjuns commented 1 year ago

Wow thanks a lot !