Inspiration
Online shopping is great but you find one piece you like but everything else on the site is not your taste, it would be great if you can find similar articles. If you see clothing woren by a model it looks good most of the time but how does it look on you?
What it does
It is a online store where you can find other articles that are similar to the one you are interested in. You can try the clothes on by uploading a picture of you
How we built it
We downloaded and preprocessed the data, after that we experimented the first idea was to use a ai to label the images by clothing type, color and design, this failed because the data we got wasnt good enough and it took too long. After that we experimented with ResNet, VGG and Xception image recognition model, there the ResNet model performed best.
In the meantime we built a frontend of a onlinestore to show the clothing products and their similar images, this is built with javascript and react in the frontend and java and python with fastapi in the backend.
we also used a image manipulation ai to show the users how they would look in the cloths they are looking at.
Challenges we ran into
The dataset was huge and took a lot of time to process and train on. We have never worked with image recognition and machine learning so it was completly new and difficult.
Accomplishments that we're proud of
Having a good looking frontend Having a functioning image similarity algorithm Having a working tool to try out clothes on your own picture
What we learned
A lot of interesting information about image recognition models and how to train them.
What's next for Dress To Impress
Built With
- java
- javascript
- python
- react
- resnet
- tenserflow
Log in or sign up for Devpost to join the conversation.