How to train your data?

[gateway]

Sorry if this is in the documentation, I have a set of jpgs that I have and ran bash scripts/local_colmap_and_resize.sh my_dataset_dir but from here im stumped as to what to do next?

Also does this train 360 equirectangular images or video?

[bmild]

You should be able to run something like

python -m train \
  --gin_configs=configs/360.gin \
  --gin_bindings="Config.data_dir = 'my_dataset_dir'" \
  --gin_bindings="Config.checkpoint_dir = 'my_dataset_dir/checkpoints'" \
  --logtostderr

[gateway]

You should be able to run something like

python -m train \
  --gin_configs=configs/360.gin \
  --gin_bindings="Config.data_dir = 'my_dataset_dir'" \
  --gin_bindings="Config.checkpoint_dir = 'my_dataset_dir/checkpoints'" \
  --logtostderr

cool and for just normal images? btw this is the 360 camera I want to be testing.. can i just use the 360 video or ? https://www.insta360.com/product/insta360-oners/1inch-360

[momentmal]

Sorry if this is in the documentation, I have a set of jpgs that I have and ran bash scripts/local_colmap_and_resize.sh my_dataset_dir but from here im stumped as to what to do next?

Same here, I ran "local_colmap_and_resize.sh" (normal images, no 360°) and don't really know what to do next. I am completely new to this and the documentation completely loses me at the end of "using your own data".

[bmild]

Hi, so far we don't support stitched 360 data but you can use fisheye data (the direct output from one half of the 360 camera) if you use the OPENCV_FISHEYE argument to the colmap script.

I just added more explicit instructions here for what to do after running local_colmap_and_resize.sh.