<aside> <img src="/icons/groups_blue.svg" alt="/icons/groups_blue.svg" width="40px" /> Team 8 Information
👩🏼🦰 이선재, Personal page
💩 전지민, Personal page
$\pi$ 정재우, Personal page
</aside>
<aside> <img src="https://s3-us-west-2.amazonaws.com/secure.notion-static.com/042da161-5032-48fb-b875-6b7271e5a91e/intersection.png" alt="https://s3-us-west-2.amazonaws.com/secure.notion-static.com/042da161-5032-48fb-b875-6b7271e5a91e/intersection.png" width="40px" /> Color Grader ↗️
</aside>
<aside> <img src="https://noticon-static.tammolo.com/dgggcrkxq/image/upload/v1566899596/noticon/slhw4nu8hybreryigopq.png" alt="https://noticon-static.tammolo.com/dgggcrkxq/image/upload/v1566899596/noticon/slhw4nu8hybreryigopq.png" width="40px" /> Github ↗️
</aside>
Nowadays, Instagram is used as a means of expressing oneself, sharing with people by uploading photos of their daily life or anything they want to share. Among them, the feed section is a long-lasting element, and many people spend time managing their Instagram feed. A common challenge faced by newcomers involves unifying their feed's color style, which may require using filters or post-processing techniques to ensure color consistency across photos. This task can be both time-consuming and difficult for beginners.
In this project, we aim to help users achieve a unified style by allowing them to provide 9 photos they wish to match the style of. When a new photo is added, its color will be automatically adjusted to fit the established style. The initial 9 photos are gathered by default through Instagram crawling. If a user doesn't have 9 photos in their feed, they can either fetch posts from other public accounts or upload their own to set the initial color style.
We call this process "Color Grading".
Fig 1. An example of well-blended Instagram feed
We chose Streamlit as our framework because it allows us to quickly create web applications without complex web development knowledge. Streamlit was a good fit because our application to create didn't require a lot of functionality or complex interaction with the user.
The structure of our application is shown in Fig. 2. We use Streamlit as our main application framework and use the Nginx web server to communicate with the client side. Additionally, the Deep Learning (DL) model is hosted on an A10 GPU, and inference takes place on the GPU when a user request is received.
Fig 2. Application Architecture
