Problem generating interpolation from Yale Faces B about deconvfaces HOT 2 CLOSED

The generate function is a really hacky thing I wrote quickly to be able to create animations, and the Yale database is really a tacked-on afterthought that I did to try to find an open dataset, so I'm really not sure what happened from this stack trace alone.

The yaml files have different parameters for Yale though -- instead of "or" (orientation) and "em" (emotion), it's "lt" (lighting) and "ps" (pose). I think I also hard-coded the number of identities in there, so if your dataset is missing people or has extra people it might crash.

from deconvfaces.

Finally got it to work.

The following is a summary of what I had to do to make it work with the Extended Yale Face Database B:

1. As mentioned above the parameters file needs to be changed to reference only valid ids
2. The Yale DB uses different parameters (as mentioned above) so the parameters file needs to be updated accordingly.
3. Earlier I trained only one identity yaleB30 which then produces all kinds of error during the generation phase. I decided to switch to consecutive identities from the top, i.e., yaleB01 , yaleB02 , and yaleB03, the errors then magically went away.
4. It seems that the variable NUM_YALE_ID in generate.py is not getting automatically updated to the total number of identities as expected. If I manually hard-code it (which should be 3 in my test case) then all's well.

Hope this helps somebody.

@zo7 Many thanks for the help.

from deconvfaces.