# 8/29/24 Meeting Notes #3 *`Final idea: AI Fashion Maker`* ## Datasets - DeepFashion : -- https://mmlab.ie.cuhk.edu.hk/projects/DeepFashion.html - Polyvore : -- https://paperswithcode.com/dataset/polyvore - we could find some datasets from the papers we found: -- Lookastic : in https://arxiv.org/pdf/2004.06229v1, they mentioned lookastic in the dataset section ## Papers we were looking some papers on these websites: -- https://ieeexplore.ieee.org -- https://paperswithcode.com we found these papers that relate the most to our project - Style analysis of Clothing from Fashion Shows Based on Deep Learning -- https://ieeexplore.ieee.org/document/9943872 - Fashion Style Recognition Using Component-Dependent Convolutional Neural Networks -- https://ieeexplore.ieee.org/document/8803622 - Dress With Style: Learning Style From Joint Deep Embedding of Clothing Styles and Body Shapes (can be added as a feature) -- https://ieeexplore.ieee.org/document/9034172 - Imitation Learning for Fashion Style Based on Hierarchical Multimodal Representation (can be added as a feature) -- https://paperswithcode.com/paper/imitation-learning-for-fashion-style-based-on - Fashion Recommendation Based on Style and Social Events (can be added as a feature) -- https://paperswithcode.com/paper/fashion-recommendation-based-on-style-and ## Features Some features that can be added into the app/website: * Main features: -- Recommends some outfits according to the style chosen using user's wardrobe. - Side features: -- Users can manually choose set(s) that they prefer. -- Acts like pinterest, where the user can post their outfit onto the internet. -- Put the AI-generated set onto a mannequin/model that has the user's body measurement (Measurements are taken beforehand). [Skin tone might be included also] -- The face of the mannequin/ model can either be the face of the user or blank like a normal mannequin. -- Users can include or exclude some clothes to be AI-generated with. Supposed you select a piece of shirt(s) that you like, then the AI will generate sets that goes with the shirt(s). -- Randomized sets. Generate random sets with all/selected clothings, wont have a specific style. -- Not only shows the generated set, but also show the user the reference of the set (similar set from the internet or from the dataset). -- Users are able to save the set that they like. -- Users are able to put clothes into likes and dislikes. ## Application We found this website that similar to our idea where you can add your own clothes to the website and find some outfits that you can create. https://lookastic.com ![image](https://hackmd.io/_uploads/BkrPKdTsC.png) - The difference: -- this website provide shopping features where you can buy items -- this website doesn't allow you to upload photos of your own clothes from your wardrobe, it has to be the clothes that they provide on their website. ![image](https://hackmd.io/_uploads/HkIk9OTsA.png) --- Previous: [8/11 Meeting Notes #2](https://hackmd.io/@emps-113up/meeting2) Next: [9/12 Meeting Notes #4](https://hackmd.io/@emps-113up/meeting4) Full Content List [here](https://hackmd.io/@emps-113up/full-list)