A request for help when testing Hpatch `ValueError: zero-size array to reduction operation minimum which has no identity` about image-matching-toolbox HOT 4 CLOSED

Hi @GabbySuwichaya !

For faster debugging, I used superpoint since it runs fastest when one has a gpu.
I ran the following command:

# Turn on --print_out to see each sample results
python -m immatch.eval_hpatches --gpu 0 \
    --config 'superpoint' --task 'both' \
    --root_dir . --print_out

And here are the console outputs:

So to figure out why you have that problem, you can do:

1. Check whether your dataset are setup correctly: it should be placed as 'data/datasets/hpatches-sequences-release'
2. Run this notebook from your local to see whether the repository is properly setup. If not, make sure you didn't miss anything in the readme.
3. Run the command i posted for hpatch, and post the output here, if 1 & 2 didn't solve your problem.

from image-matching-toolbox.

Thanks!
I found the error now, and the problem is solved. It was the directly of the HPatches data...

Here I would like to kindly ask one more question...
There is a line which said Cannot import sparsencnet ...

Could you please guide me what causes this line... ? I am very sorry I am not very familiar with the code...
Also should I try to solve it?

from image-matching-toolbox.

Hi @GabbySuwichaya ，

That line will appear when your environment doesn't contain MinkowskiEngine which is required by SparseNCNet. This is explained in install.md. So don't worry about it. It's expected to be there unless you want to run SparseNcNet, for which you need to follow the install.md.

from image-matching-toolbox.

Hi @GrumpyZhou,

Thanks so much for this. Yes, I will fix the problem accordingly. :)

from image-matching-toolbox.