Skip to content

zitansmail/CBIR-New-Search-Engine-Approach

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

44 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Open Source Love

Intro

This repository contains a CBIR (content-based image retrieval) system Implimentation for my conference paper

Full article

Extract query image's feature, and retrieve similar ones from image database

Resume

In this system our approach is based on the ViT architecture for ex- tracting features from the Corel images. Then, we used the PCA as a features selector to minimize the dimensionality. Finally, we implemented the Annoy algorithm for similarity searches.

Feature Extraction

Dimension Reduction

The curse of dimensionality told that vectors in high dimension will sometimes lose distance property

  • PCA

Evaluation

CBIR system retrieves images based on feature similarity

we have evaluated our CBIR system using the Cored-1K dataset with the same ten aforementioned classes; then, the precision and recall metrics are computed for each category, individually and overall. These metrics are calculated based on the top 20 images retrieved. Table below shows the findings results.

Category Precision (%) Recall (%)
Buses 100 20
Mountains 100 20
Beach 90 18
Elephants 100 20
Food 100 20
Flowers 100 20
Africa 100 20
Horses 100 20
Dinosaurs 100 20
Buildings 100 20
Average 99 19.8

Author

Zitane Smail / @zitansmail

Releases

No releases published

Packages

No packages published

Languages