Image Search using Deep Learning

Content based Image Retrieval (CBIR) is a popular technique used in computer vision for finding out a specific image out of a given set of images. Different category of images is stored in a folder of the repository. These are then preprocessed and visualized using image processing technique. Autoencoder architecture is used for image classification. Deep learning model is trained with the train set of data while model inferencing is done by using test component of dataset. When the user wants to search a particular image which will be matched with each of the images present in the repository and the image that is best matched with the given image will be displayed on the top followed by other similar images based the similarity score. Flask API is used to develop a web application running on local server. This project finds many useful applications such as fashion and retail, manufacturing and health care industries.

Salient Features:
– Image processing for proper visualization
– Categorization of images
– Applying Autoencoder for content based image retrieval
– Using similarity search algorithm to find the top 5 similar images
– Simple UI for searching an image

Tech Stack:
Deep Learning Frameworks, OpenCV, CNN, Python, Flask, HTML, CSS

Industry:
Fashion, Retain, Manufacturing, Healthcare and Life Sciences