norouzi/mih

MIH runtime performance with not so big database

inspirit opened this issue · 2 comments

inspirit commented

Hello,

i was testing MIH matching 500 query codes of length 293 Bits to 10K-20K in database.
my setup for parameter "m" (number of hash tables) is 19 to 14. KNN is always 2.

the matching results are good but runtime performance is worse compared to brute force matching each query to everything in database.

Can you please give me any ideas on how to improve runtime performance?

cbalint13 commented

@inspirit,

  • Agree with that. Something is odd with latest commits from this year, 2013 code runs better.
  • 👍1
danielcrane commented

Same here.

@cbalint13 Doesn't the 2013 code require MATLAB in order to be able to run it, though? :(