--------Relja Arandjelović and Andrew Zisserman (CVPR 2012)

1. RootSIFT:

 Improves performance in every single experiment (not just retrieval);

 Every system which uses SIFT is ready to use RootSIFT;

 Easy to implement, no added computational or storage cost;

2. Discriminative query expansion:

 Consistently outperforms average query expansion;

 At least as efficient as average QE;

 No arguments against it except for slightly increased implementation ;

complexity

3. Database-size feature augmentation:

 Useful for increasing recall;

 Our extension improves precision but increases storage requirements; this trade-off should be considered when deciding whether to use it or not;