Image-Search-System-Using-CLIP-Model

The project is completely developed on the aws platform, and it is backed up on github for easy sharing.

Introduction

The CLIP model is an image and text matching model developed by openAI, which can be used for multiple purposes. This project implements a simple image search system implemented by matching text and image feature vectors, supporting image search and image upload.

Usage

Back-end

--Deploy the docker image 'public.ecr.aws/d2x9y6a2/wy-usage:1.0' on AWS ECS.

--Copy lambda function 'amplify/backend/fucntion/clipimage/src/index.py' and 'amplify/backend/fucntion/textimage/src/index.py' for creating your lambda functions of python3.9, replace the api_url with the ip address and port of your own deployed container.

--Add a layer containing package numpy and requests for lambda functions

--Configure dynamoDB and S3 and replace them in lambda functions.

--Create your own api for lambda functions

Front-end

--If you use amplify, modify the api name and path in the page clip.js to your own.

--If you don't use amplify, you can write your own function to call your api

--npm intall

--npm start