Week #6 #

Deployment: styleU #

Design: figma #

1. Final deliverables #

Project overview #

Our project is an AI stylist that helps a person choose looks for events and returns a set of things with links where they can be ordered, while it takes into account the cost of goods, as well as the features of a person’s figure and color type - therefore, it chooses not just trendy goods, but also what will suit a person personally

Features #

color type definition by photo + personal recommendations from AI
body type definition by parameters + clothing recommendations
image selection for an event based on the user’s personal request - all you have to do is follow the link and order

Tech stack #

ML: numpy, PIL, skimage, sklearn, cv2, mediapipe, torch, clip
Backend: python, mongo, posgres, fastAPI
Frontend: Jsx(vite)+react+TailwindCSS

Setup instructions #

clone repo
create .env file by example - we can send you all passwords
run docker-compose up

go to our deployed version & try it

2. Presentation draft #

We have only plan for presentation:

Slide about marketplace statistics + popularity of rice with clothing reviews => decided to study this topic
Slide about the target audience
Slide about the fact that many people, as it turned out, have problems with choosing clothes (here show graphs from our survey) and highlight a list of reasons
- Clothes are expensive
- <….>
- They don’t know their color type
Here it’s worth telling in more detail about the color type - we provide a diagram where the difference in color types is visible + a link to an article and a video from YouTube2 slide show what color types look like (cut photos from Pinterest)
Conclusion from all this - we decided to make an AI stylist who will determine the body type, color type and recommend images made up of real products from popular marketplaces with the ability to see how these things will look on a person
Here you can slide about the team and about the fact that they worked on an agile type

Slide about feature 1 color type - an example of how it works, why it is useful for the user, they can think about buying clothes in colors that suit them or just compare that things in these shades will suit them less
- Slide about ML - how this thing works in general with an example of a graph
Slide about feature 2 figure - an example of how it works and why it is useful - like we tell about the features of the figure, throw in ideas about what suits and what doesn’t + then we use this feature to select images, also tell here in 1-2 sentences how it works
Slide about feature 3 image selection
- An example of how it works from the user’s side, like entered data received an image
- Explanation of what lies behind it - a diagram + tell about the parser
Slide about the avatar feature - clarify that the idea came from a trend on Instagram (this is close to our target audience) and then it may develop into a 3D model, for now it’s just a picture to see how things go together and send to a friend??

Now about how it all works inside

Alena will talk about the general architecture of the product and delve into the back- Alena will also talk about the database and authorization
Aisylu will tell what model we use and how it all works
Katya and Masha will talk about the front and design
- How we chose the design, about the survey, etc.
- About the architecture of the front and what we used for writing, you can talk about the translation here
Life demonstration