An AI-based image search engine that includes deep transfer learning features Extraction

Deep Image Search

Deep Image Search is an AI-based image search engine that includes deep transfer learning features Extraction and tree-based vectorized search technique.

Features

Faster Search O(logN) Complexity.
High Accurate Output Result.
Best for Implementing on python based web application or APIs.
Best implementation for College students and freshers for project creation.
Applications are Images based E-commerce recommendation, Social media and other image-based platforms that want to implement image recommendation and search.

Installation

This library is compatible with both windows and Linux system you can just use PIP command to install this library on your system:

pip install DeepImageSearch

If you are facing any VS C++ 14 related issue in windows during installation, kindly refer to following solution: Pip error: Microsoft Visual C++ 14.0 is required

How To Use?

We have provided the Demo folder under the GitHub repository, you can find the example in both .py and .ipynb file. Following are the ideal flow of the code:

1. Importing the Important Classes

There are three important classes you need to load LoadData - for data loading, Index - for indexing the images to database/folder, SearchImage - For searching and Plotting the images

# Importing the proper classes
from DeepImageSearch import Index,LoadData,SearchImage

2. Loading the Images Data

For loading the images data we need to use the LoadData object, from there we can import images from the CSV file and Single/Multiple Folders.

# load the Images from the Folder (You can also import data from multiple folders in python list type)
image_list = LoadData().from_folder(['images','wiki-images'])
# Load data from CSV file
image_list = LoadData().from_csv(csv_file_path='your_csv_file.csv',images_column_name='column_name)

3. Indexing and Saving The File in Local Folder

For faster retrieval we are using tree-based indexing techniques for Images features, So for that, we need to store meta-information on the local path [meta-data-files/] folder.

# For Faster Serching we need to index Data first, After Indexing all the meta data stored on the local path
Index(image_list).Start()

3. Searching

Searching operation is performed by the following method:

# for searching, you need to give the image path and the number of the similar image you want
SearchImage().get_similar_images(image_path=image_list[0],number_of_images=5)

you can also plot some similar images for viewing purpose by following the code method:

# If you want to plot similar images you can use this method, It will plot 16 most similar images from the data index
SearchImage().plot_similar_images(image_path = image_list[0])

Complete Code

# Importing the proper classes
from DeepImageSearch import Index,LoadData,SearchImage
# load the Images from the Folder (You can also import data from multiple folder in python list type)
image_list = LoadData().from_folder(['images','wiki-images'])
# For Faster Serching we need to index Data first, After Indexing all the meta data stored on the local path
Index(image_list).Start()
# for searching you need to give the image path and the number of similar image you want
SearchImage().get_similar_images(image_path=image_list[0],number_of_images=5)
# If you want to plot similar images the you can use this method, It will plot 16 most similar images from the data index
SearchImage().plot_similar_images(image_path = image_list[0])