Clovet

Inspiration

Clovet was born from a simple observation: we all stand in front of overflowing closets feeling like we have nothing to wear. This paradox isn't just frustrating – it's a $460 billion crisis. People forget 50% of their wardrobe, wear each garment only 7 times before disposal, yet still feel compelled to buy more. We were inspired by the disconnect between intention and action in sustainable fashion. 60% of consumers want to shop secondhand, but fragmented platforms make it feel like a part-time job. We realized the fashion industry doesn't need more clothes – it needs better tools to help people wear what they already own and shop sustainably when they genuinely need something new.

What it does

Clovet is a wardrobe assistant that encourages sustainable fashion habits through three core features:

Digital Wardrobe Cataloging - Users upload photos of their clothes or paste product links. Clovet automatically organizes everything by type, color, style etc., making their entire closet searchable.
Unified Secondhand Search - One search query pulls listings from multiple secondhand platforms (Depop, Poshmark, ThredUp, Carousell, and more) simultaneously. No more tab-hopping across dozens of sites. Our secondhand search also employs semantic search language so you are able to search based on vibes ("coastal granddaughter core", for example) just as easily.
AI Virtual Try-On - Powered by Gemini AI, users can see how secondhand items look on their body before purchasing, reducing returns and building confidence.

The platform will also provide personalized outfit recommendations based on existing wardrobe items.

How we built it

Frontend: React.js

Backend: Node.js

Database: MongoDB

AI: Gemini AI for virtual try-on and realistic garment visualization

ML: CNN machine learning model for wardrobe feature extraction and image classification

Marketplace Integration: Carousell API for secondhand listings (more API integrations to come!)

Challenges we ran into

1. Lack of Readily Available Marketplace APIs:

Major North American resale platforms (eBay, Etsy, Poshmark, Depop) either don't offer public APIs or require 1-2 business day approval processes for developer accounts. To maintain momentum, we pivoted to Carousell, a Singaporean marketplace with more accessible developer tools. This worked for our proof-of-concept but highlighted the need for strategic partnerships as we scale.

2. Processing Speed and Scalability:

Running CNN feature extraction on images is computationally expensive. Processing large wardrobe uploads (50+ items) can take significant time, which impacts user experience. Balancing accuracy with speed remains an ongoing challenge as we scale.

Accomplishments that we're proud of

Built a functional product in 24 hours
Created beautiful, intuitive UI/UX
Developed a product that aims to encourage sustainable fashion habits
Successfully integrated Gemini AI for virtual try-on
Overcame API limitations with creative problem-solving

What we learned

Training models for fashion classification is uniquely challenging because of subjective interpretation and context-dependent styling. A "vintage blazer" means different things to different people, and the same item can serve multiple purposes.
The secondhand marketplace ecosystem is surprisingly closed. Despite the resale boom, many platforms guard their data closely. Building relationships and partnerships will be as important as technical development.

What's next for Clovet

Secure API partnerships with major secondhand platforms (Depop, Poshmark, ThredUp, Vinted)
Refine model with real user data and expand training dataset for improved accuracy
Potential future features:
Carbon footprint visualization showing environmental impact of wardrobe choices
B2B offerings for circular fashion brands to connect with conscious consumers