Content Based Image Retrieval System (CBIR)

This is an attempt to build a CBIR system for images. You can use your own dataset of images for which you wish to build a CBIR system.

Requirements

• Keras
• Matplotlib
• sklearn

Working

The system uses a Convolutional Neural Network (ResNet-50) to extract the features of all the images in the dataset. Then these features are clustered by k-means clustering (supervised, meaning you know the number of different classes of images in the dataset. Finally, any number of similar images can be obtained using a query image.

To use this system:

• Copy the dataset of all the images (all images belonging to different classes in single directory).
• Use Config.txt to configure the path settings for dataset, clustered data, and models.
• run Cluster_Data.py to extract features of all images and cluster the images based on the extracted features.
• run Search_Images.py to retrieve desired number of similar images.