# Bag of Words Image Retrival
![[visual-bow.jpg]]
Naive approach based on [[Image Feature Representations#Bag of visual words]].
- Offline: construct BoW Vocabulary
- Offline: All images in the dataset -> BoWs
- Create description from Query image
- Compare to all images in the dataset
- Rank according to similarity
Evaluation
- [[Classification Metrics and Evaluation#Precision and Recall]]
- Average precision is also used: $\mathrm{AP}=\frac{\sum_{r=1}^{N}(P(r) \times \operatorname{rel}(r))}{\text { number of relevant documents }}$
---
## References