How to retrieve images by visual similarity using deep lear
- Artificial Intelligence
- Computer Vision
- Project length: 0h 43m
This tutorial will cover the basics to learn how to find the most visually similar images from a set, to a query one, using the python language over a unix operating system. You will be guided through the process of setting a minimal environment and implement the scripts to achieve this objective. You will learn how to use a pre-trained deep model from the Keras library to generate a vector representation of each image. This vector representation will then be used to find similar images to a query one.
What are the requirements?
- Basics of python
- Unix operating system
- Basics of unix command line usage
- Virtualenv tool of unix pre installed
What is the target audience?
- Learners who want to start learning the basics of using deep learned models
- Learners who want to enhance their knowledge in Artificial Intelligence
- This course will help learners to make the first steps towards the implementation of an image search engine
Session 1: Setting-up the Environment.
Here we will be creating a virtual environment to encapsulate the package installation and Installing the python packages needed
Session 2: Basics of deep learning.
Here we will learn What is deep learning? Also What is The VGG16 Neural Network and how to use the VGG16 Neural Network to characterize images
Session 3: Implementing a python class to characterize images.
Here we will learn how to Importing the needed packages, preprocessing images using Keras and using a pre-trained Keras Deep Learned Model to get a vectorized representation of an image.
Session 4: Implementing a python script to index a set of images to a csv file.
Here we will learn how to read and characterize the set of images and save the information to a csv file as a kind of database.
Session 5: Implementing a python script to retrieve similar images to a query one.
Here we will learn how to read and characterize the query image, read the csv database file, find similar images by distance, and plotting the retrieved images using pyplot.