Ai OOTD banner

Ai OOTD

3 devlogs
17h 19m 45s

Can’t decide what to wear? Just upload your clothes and get ideas from AI!

This project uses AI

Used to generate OOTD ideas

Demo Repository

Loading README...

Liz

Shipped this project!

Hours: 17.33
Cookies: 🍪 132
Multiplier: 7.62 cookies/hr

I built my first website from scratch! It helps people with decidophobia to pick their OOTD. Through the process of building, I learnt new language: HTML, CSS and Javascript. I also learnt about types of AI, LLM model, computer vision model and CDN hosted AI which runs locally without API. Looking back, I would never believe myself that I could create a website with AI from scratch!! I’m so happy of this project :>

Liz

FINAL TOUCH UP
I found out I can make it a static web which run only with front end instead of creating a backend server. This make the website much more simpler and free to deploy (I deploy it using github), while ensuring privacy of user.

Fixed several bugs + fallback: button loading not displaying, show message when image failed to load instead of silently failed, fallback if AI failed to load, fallback if keywords failed to extract
Added comments for code understanding.
Result:
(I’m using dark mode, default it’s light theme)

0
Liz

I continued by designing and decorating the webpage. To start with, I scroll through Pinterest and other digital wardrobe website to get inspiration. I want to make the webpage simple, modern and aesthetic, most importantly, user-friendly. After that, I learned HTML and CSS to build the frontend with youtube and claude ai for practices and further understanding.

To make my webpage functional, I used javascript to for the logic. However, I was fustrated because whenever I refresh the page, my data and items are lost. After doing several hours of research, I figured out how to upload and store files at local storage.
That’s what I have work so far. More updates coming soon. :>

(fun fact: It’s my first time creating a webpage).

Attachment
Attachment
0
Liz

I went searching for models that can understand images and turn it into words for further analysing work. I found CLIP, a model which pair image with the labels that best describe the image. I also used some libraries for data handling.

Attachment
0