What is the best index type for face recognition tasks? vector length 512. What is the best recall rate/speed compromise #34675
Unanswered
martinenkoEduard
asked this question in
Q&A and General discussion
Replies: 1 comment 3 replies
-
|
Maybe someone tried to solve similar problem? |
Beta Was this translation helpful? Give feedback.
3 replies
-
|
Usually HNSW is a good choice. You can still get 99% recall or above on top10 retrieval |
Beta Was this translation helpful? Give feedback.
-
What are good starting parameters of HNSW |
Beta Was this translation helpful? Give feedback.
-
|
you don't need to setup any parameter for hnsw |
Beta Was this translation helpful? Give feedback.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
I have made a software for face recognition.
currently I use FLAT index. Because it has 100% recall rate
But after reaching 10 millions entries it start working not so fast.
what is the best index for this purpose?
Like I need close to 100% recall rate especially really close neighbors.
Like I don't really care about vectors that are 100th in the similarity, but I really care about first 10 found entities.
Beta Was this translation helpful? Give feedback.
All reactions