🔍 About our Project:
The visual search engine retrieves videos or images based on descriptive captions representing the visuals. The captions are given as a search keyword to point to the scenario. This project can be deployed for real-world scenarios such as searching for a particular incident from security cameras for investigations, traffic rule violations, personal video organizing, searching for moments from long-duration videos, and searching for images with captions.
❓ The Need of our project:
We chose this project to meet real-world scenarios such as personalizing media, retrieving the required image from each one’s gallery, or a backup drive while glimpsing old memories or requiring important information and providing security for traffic management, adhering to traffic rules. It is very time-consuming to manually review video footage for retrieving information. The visual search engine processes the caption provided according to users’ needs and gives the desired output.
⚙️ Tools Used:
Front-end: React JS,HTML, CSS, Java Script
Backend: Python | Embedding Models -OpenAI CLIP/ Sentnace Transformer clip-ViT-B-32-multilingual-v1
Database: Vector DB
⚓ Libraries used:
*FFMPEG
*Open CV
*Pillow
*OpenAI CLIP
*S3 Bucket
*Flask
👨🔬 TEAM MEMBERS:
Alisha Thomas | Aswathy Anilkumar | Jasmin Joseph | Sudhy Sukumaran