Skip to content

Can we design an effective image retrieval system using Deep Learning

Notifications You must be signed in to change notification settings

lulmer/deep_image_retrieval

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 
 
 

Repository files navigation

deep_image_retrieval

Louis Ulmer - Data Scientist

Presentation

An AI based Image Retrieval System In this project we will build an information retrieval system. We have a query image and we want to retrieve the most visually similar image for a given dataset. The main idea is to extract relevant vectorial features from images in order to measure a meaningfull similarity distance.

How to use it

Consultation Only

The demo.iypnb file is a notebook that contains all the workflow for the project, with details, comments and results, therefore you can use it as an example and reuse some pieces of codes in your pipelines scripts.

Executing the notebook

If you are interrest in executing this demo.iypnb notebook and reproduce the results.

  • 1- Fork the repository
  • 2- Download the dataset (link is in the notebook)
  • 3- Replace the path of the import cell (commented on the notebook)

References

https://en.wikipedia.org/wiki/Content-based_image_retrieval https://towardsdatascience.com/bag-of-visual-words-in-a-nutshell-9ceea97ce0fb https://towardsdatascience.com/review-resnet-winner-of-ilsvrc-2015-image-classification-localization-detection-e39402bfa5d8

One shot learning consists in learning from one (or a few) examples. It is a concept closely related to image retrieval. If you are interested in siamese networks for image retrieval, you can read these 3 blog posts, which mostly focus about one shot learning: https://hackernoon.com/one-shot-learning-with-siamese-networks-in-pytorch-8ddaab10340e https://sorenbouma.github.io/blog/oneshot/ https://towardsdatascience.com/one-shot-learning-face-recognition-using-siamese-neural-network-a13dcf739e

About

Can we design an effective image retrieval system using Deep Learning

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published