The Virtual Dressing Room

Team MindWave's Dressing Room revolutionizes the TikTok Shop experience by personalizing fashion searches. Users can add items to a wishlist and access a virtual canvas, matching outfits effortlessly.

Comment 3

Inspiration

The inspiration for MindWave came from observing how overwhelming the TikTok Shop journey can be for users. With countless options and endless scrolling, finding the perfect fashion items can feel like a daunting task. We wanted to create a solution that simplifies this process, making it not only more efficient but also enjoyable. By leveraging advanced technology to hyperpersonalise the shopping experience, we aimed to transform how users discover and interact with fashion products on TikTok Shop.

What it does

MindWave revolutionises the TikTok Shop experience by providing a hyperpersonalised search for fashion products. Users can seamlessly add items to their wishlist and later access them in a virtual dressing room. This feature allows users to mix and match outfits effortlessly, eliminating the need to compare multiple product pages. The virtual dressing room is easily accessible via a pop-up button on the 'For You' page when fashion-related ads appear or from any virtual shop page within TikTok Shop. From the dressing room, users can navigate to their wardrobe page, which includes clothes they personally own and items on their wishlist. MindWave offers a tailored, intuitive, and engaging way for users to discover their ideal fashion products.

Features Offered

  • Users are able to upload pictures of their clothes. This should be done with their clothes placed on a flat surface. With the use of sophisticated Machine Learning models, these images are tagged and the background for the clothes are removed. The use of tagging allows for searchability, while the removal of image backgrounds improves the styling experience. These processed images are then stored in a database and only accessible to the user via the User Wardrobe.

  • Likewise, businesses and shops are also able to upload images of clothing items to their very own Shop Wardrobe. These images are available to users, who maintain a User Wishlist for styling purposes before purchase.

  • The standout feature of MindWave is the virtual dressing room, offering users a blank canvas to style themselves. This feature is conveniently accessible via a pop-up button on the 'For You' page when fashion-related ads appear, or from any virtual shop page within TikTok Shop. In the virtual dressing room, users can mix and match outfits effortlessly, combining items from their wardrobe and wishlist to find their perfect fit.

  • Lastly, we also offer a suggestions feature, which provides a list of recommended outfits that complement a selected piece of clothing. This feature makes use of colour theory, and our recommendation system looks out for other clothing pieces whose colour palette are similar to the current piece of clothing, or forms its complementary colour palette.

How we built it

We built MindWave by integrating advanced machine learning models to analyse user preferences and behaviors. Our models are used to remove background images from product photos, creating a clean and cohesive browsing experience. We also developed a suggestions feature to help users discover new items, and implemented a tagging system using the Fashion-CLIP model to classify clothing into different types. Our team worked tirelessly on user interface (UI) and user experience (UX) design to ensure that the platform is intuitive and user-friendly.

Development tools

Frontend:

  • React.js
  • Fabric.js
  • Chakra UI + Tailwind CSS
  • Vite

Backend/ML:

Challenges we ran into

One of the main challenges we faced was ensuring the accuracy and reliability of the machine learning models, as well as the algorithms we devised for our tasks. Fashion-CLIP can be unreliable if the input classes were too general. Therefore, we had to find new ways for integrating the CLIP model for accurate clothing classification presented technical difficulties, particularly in maintaining consistent and precise tagging. We also had a lot of difficulty devising an algorithm for recommending clothes in one's wardrobe. The frontend also encountered many challenges too, particularly in devising the scrolling effects of the reels, as well as dealing with conflicts that occur between the libraries. For example, ChakraUI has its own predefined classes, whose effects come into conflict with other classes. For more details, please visit our Github repository, and for specific details, please refer to our repository's docs folder.

Accomplishments that we're proud of

We are proud of successfully creating a platform that significantly enhances the TikTok Shop experience. Our hyperpersonalisation algorithms have demonstrated a lot of usefulness in recommending fashion products that match users' tastes and preferences due to the myriad of features it provides. We also take pride in our user-friendly interface, which can make the shopping and styling experience more enjoyable and efficient for users.

What we learned

Throughout the development of MindWave, we learned the importance of understanding user needs and preferences. Continuous feedback from users was crucial in refining our algorithms and improving the overall user experience. We also gained valuable insights into the technical aspects of integrating AI technology into our application. The project taught us the significance of collaboration and iterative development, as working closely as a team and constantly iterating on our design and features led to a successful outcome.

Learning about AI integration into software was also a very interesting experience and provided opportunities for team members, who are proficient in a wide variety of skills, to share their varying ideas.

What's next for MindWave

Moving forward, we plan to enhance MindWave by incorporating more advanced features and expanding its capabilities. We envision possible plans to integrate social sharing features, allowing users to share their favourite outfits and receive feedback from friends. We also forsee efforts to explore partnerships with fashion brands to offer exclusive deals and promotions to our users.

Built With

+ 46 more
Share this project:

Updates