Faster Fashion

Imagine you see someone wearing a really nice outfit and want to find similar-looking pieces for yourself. With Faster Fashion, you can build a new wardrobe all with the use of an image.

Comment 2

Inspiration

We felt that sometimes when coming across an article of clothing that we like, it's hard to find a wide assortment of similar clothes to select the best one. With that in mind, we built Faster Fashion to find and display clothes that would cater to the user's specific interests.

What it does

The user uploads an image to the program with the specific article or articles of clothing in the frame and is shown an assortment of related clothes based on features like color and type.

How we built it

We first used PostgreSQL to create the database to store the data of thousands of clothes from retailers like H&M through web scraping scripts. We then implemented the Google Cloud Vision API with Python to create an algorithm that detects articles of clothing and their colors within a given image. Using HTML/CSS, Bootstrap, Jinja, Flask, and JavaScript, we then designed a full-stack website that allows users to submit images of clothing and where we provide similar clothing through cards describing the item, its cost, an image of it, and a link to its storefront.

Challenges we ran into

  • Getting the model to properly recognize the different attributes of each article of clothing
  • Learning SQL and implementing the database
  • We were a three-man team for the last 36 hours of the hackathon because our teammate got food poisoning.

Accomplishments that we're proud of

  • Having a functional model that detects labels in images and correlates them with the user's image
  • Learning how to use new technologies like Google Cloud API and PostgreSQL

  • Creating a well-designed, easily accessible website

    What we learned

  • Programming a large-scale project is difficult and can be frustrating

  • Complex projects require many moving parts and putting them all together requires lots of planning

    What's Next for Faster Fashion

We hope to make the database available in the cloud so that the program can be accessed anywhere at any time. We also hope to have a more refined model that is able to make sharper detections in images.

Share this project:

Updates