-
This is our front-end
The Inspiration
We watch k-pop. We sometimes want to dress like them. We don't know how to pair clothing properly. And frankly, as university students, we don't have time in the morning to do it.
What it does
It gives you recommendations on how to dress. You select your desired dress style and input your clothes and it'll pump out different ways to pair them.
How it's built
Front-end built with HTML, CSS, JS/jQuery. Connected front-end with back-end using Django. Used IBM Watson's computer vision API to recognize the types of clothing seen in pictures. Used Python to conduct internet scrapes to identify what type of clothing combinations are trendy. Used Machine Learning to use that information to generate fashionable combinations of clothes.
Challenges we ran into
When conducting scrapes to train our machine learning algorithm, there was a lot of irrelevant data. We fixed it by scraping with key words so that the photos get filtered properly. Making a database with Django was hard (since we had little to no Django/Python knowledge). Sending scraped photos to the database was hard. When we connected our front-end to our server, the website didn't appear properly so we had to edit the Python code to make it appear properly. Also, downloading Django through command line is a pain.
Accomplishments that we're proud
Implemented error handling with Watson IBM API, filtering the testing data based on specific keywords, and developing IBM custom models with different classes to determine the weights between each image being processed.
What we learned
Gained experience on using Django. Scraping by keyword and making databases. Knowing what’s a good testing data for the Watson IBM Watson API visual recognition, creating a basic foundation for a custom model, learn how to retain model based on new photos, using IBM Watson API to notice trends in fashion
What's next for HypeGoose Inc.
Connect backend with frontend. Use this project to get a coop job.
Log in or sign up for Devpost to join the conversation.