--------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;