fashionably was born out of the need to find a way to speed up our mornings and bring a new degree of organization to our fashion. As first-year university students adjusting to all of our newfound responsibilities, being able to wear whatever has become a little more difficult than we imagined. We no longer have our parents reminding us to dress for the weather, or the time to plan out intricate outfits in the morning when we're trying to squeeze in those extra minutes of sleep. This became the inspiration to build fashionably; our solution to these issues and our first step to becoming efficiently independent!
What it does
fashionably is an iOS application that recommends what outfit you should wear (e.g. tops, bottoms, shoes, and outerwear if needed) depending on the weather outside (e.g. temperature, precipitation) and occasion (e.g. casual, formal). The app uses machine learning to eventually discover your style preferences and offer personalized outfit suggestions to you.
How we built it
fashionably was built mainly using Swift as well as Keras, coreML, XCode and Python.
Challenges we ran into
Developing fashionably definitely wasn't easy, and there were a couple of major hurdles that we faced along the way. From training our machine how to analyze user-submitted images successfully to figuring out a system to categorize clothing effectively, our team worked collaboratively to think through our struggles.
Two significant roadblocks that we overcame:
- Training a convolutional neural-network for this large dataset is naturally tough. Since the dataset was so large and broad, it was a challenge to utilize for our purposes.
- The clothes selection screen had a mix and match selection feature where the user could select all of their clothes separately. This was a major challenge that we ran into during the development of the GUI.
Accomplishments that we're proud of
fashionably's journey to fruition was very exciting. One of the accomplishments we're very proud of as a team was our application's overall UI/UX design. We designed the program from the ground-up with the ease of use as our utmost priority. Although there were some slight difficulties with adjustments required when converting to Storyboard, we were able to adapt. In the end, we were very happy with the balance of functionality and looks that fashionably contained, and consider it one of our greatest successes!
What we learned
We learned that ML definitely comes with a lot of its own quirks and requires some flexibility.
What's next for fashionably
There's so much more we have planned for fashionably! There are many critical features and extension we believe would greatly improve the user experience of the app, and would love to add. These include a parent-friendly version (for parents trying to choose outfits for their children), a travel planning extension (which allows a user to be able to see what to pack for an upcoming trip based on the forecasted weather), Android and website development and much more.
Log in or sign up for Devpost to join the conversation.