Made with Supabase

Lookout 👀

Lookout is an application that lets you discover images from across various media by searching them using their metadata - their properties, content inside the image, textual data found inside the image etc. It gives us a whole new take on looking and searching through images, blazingly fast! You can also upload you own set of images to be processed and indexed, made available for the public to search

Demo URL: Demo

Youtube Promo: Promo

Our workflow

We have always believed in great engineering and splendid design processes. We thought it would be cool to build a system that can take in image data and deliver a neat searching interface for the same using metadata scraped within the image. To achive this, we use almost every supabase service provided - Auth, Database, ‼️ STORAGE ‼️ and Realtime and Functions, coupled with Google's Cloud Vision APIs. We were able to scale our application to support upto hundreds of thousands of images found across public media.

Tech Stack

We use the following stack to achieve this:

  • TailwindCSS 💅🏽
  • NextJS ⚛
  • Flutter 📱
  • Supabase Services ⚙️
  • Typesense Search 🔎
  • Google Vision API 🌆

Product Architecture

We have our client applications - Web and Mobile that use client SDKs of Supabase and Typesense to fetch and post. The web uses an instant-search-adapter to provide a search-as-you-type experience that is unparalleled - hundrend thousands of results queried in sub 10ms. We use Supabase Auth to provide a login flow to users who want to upload their data. Every image is stored in supabase storage. Once a user uploads their image, we trigger an edge function that pushes the image on to Google's cloud vision pipeline and we feed the results back to Supabase (upserting the data) and to Typesense to index the search. We use supabase realtime to give the user realtime update of the progress in uploading, processing and indexing.

About us

Aman

Nivas

Hari

Sabesh

We are a team of 4 university seniors from India 🇮🇳 striving to build the best software we can. We had a wonderful ride while building this product, and will continue to be fascinated by how it turned out. We have put in our best practices when it comes to systems and product design. Cheers ✌🏽

Related Projects


A project by Zernonia