Real world problem 101 - People have been spending too much time on choosing attire of the day (excluding programmer because we always dress the same). Often time, choosing OOTD can be so disastrous that one have to dig out the entire closet to match their moods with the best outfit. Dressing in the wrong FASHION can end up being judged by the passerby - how pity. On a side note, according to statistics, average person buys 60% more clothing per year, which ends up in severe clothes wastage. In fact, human can be so forgetful, forget that the wardrobe is over-occupied, when the shopping mood hits. This project aims to better manage the fashion inventories, to better realize the value of each dress and to bring out the best appearance that one could ever be.


FASHIONITY consists of three components:

a) The Chamberlain (Clothes folding & sorting machine)

  • Automatic folding and storing of clothes in wardrobe.

b) The Proprio (Virtual Mirror - substituted by computer for demo)

  • Hand gesture controlled image capturing of your dressing
  • Analyze and classify the dressing, hence updating the database of clothes collection

c) The Window (Web app.)

  • Display all collections of clothes, dressing matching
  • Virtual clothes testing to handle all the mess

How we built it

a) Hardware part:

Due to time limit, cardboard is chosen to build the wardrobe. Arduino Uno with servo motors are programmed to automatically fold clothes and store them in the closet.

b) Software part:

  • Cloud ML -> Algorithmia Machine Learning API is deployed to detect fashion and classify the dressing.
  • OpenCV with Python is used to detect hand gesture to trigger image capturing.
  • Nodejs, with Express and Pug to build web app.

Challenges we ran into

  • Insufficient materials, so we use cardboard and recycle can to build our hardware frame.
  • Lack of appropriate tools to sort out the hardware with ease, and lack of easier ideas for demo purpose.
  • To analyse image of dressings and identify categories of clothing. Difficulties especially increase when fashion comes in numerous combinations.

Accomplishments that we are proud of

  • Manage to build a decent clothes folding machine.
  • Able to detect and categorize clothing.
  • Able to update database of each category of clothing.
  • The framework of the project is realized, showing the interrelations of each separated components and how each of them tie to achieve a foreseeable prototype.

What we learned

  • Utilize limited resources and knowledge to partially complete a multi-discipline project
  • Appreciate the open source community that makes technologies more approachable
  • Good teamwork halves the hustle

What's next for

  • Real-time tracking of clothes in wardrobe, for easier storing and retrieving.
  • Clothes classification of folding machine
  • Support choosing of clothes via app and auto-retrieving of clothes.
  • Suggest appropriate dressing based on weather, mood and freshness.
  • To receive feedback through social media for the outfit of the day.
  • AR clothes testing on a one-sided mirror.
  • Suggest donation of old clothes
Share this project: