valeoai / LOST

Pytorch implementation of LOST unsupervised object discovery method
Other
234 stars 26 forks source link

Question regarding symmetric matrix in patch_scoring function in object_discovery.py #18

Closed safdarzareef closed 2 months ago

safdarzareef commented 2 months ago

Hello,

Thank you for providing the code for the paper!

In the object_discovery file line 72 (https://github.com/valeoai/LOST/blob/master/object_discovery.py#L72):

After making the matrix A symmetric it is not used anywhere. Should this new symmetric matrix C be used instead of A in object_discovery.py#L75?

safdarzareef commented 2 months ago

Sorry, I think it doesn't matter because A is already symmetric due to the dot product before in the lost() function.