TryMe is a Chrome extension that lets users virtually try on clothing directly on any shopping website.
Powered by the IDM-VTON deep learning model (via Hugging Face), the extension allows users to:
- Paste or upload a garment image
- Paste/upload/snap a selfie
- Generate a virtual try-on preview
- Reverse-search the garment on Google Lens
- Use drag-to-move UI and live camera integration
TryMe was originally inspired by my team's NatHacks 2025 hackathon project.
The goal was to create a lightweight, virtual try-on tool to reduce clothing waste while shopping online.
Check out the original project:
Live website: Emergrade on Render
GitHub repo: Emergrade GitHub Repo
Devpost: Devpost Page Here
- Clone the repo
git clone https://github.com/JeremelleV/TryMe- Open Chrome → Extensions → enable Developer Mode
- Click "Load unpacked"
- Select this repository folder
- The TryMe panel will appear on the right side of any webpage
- Shadow DOM UI
- Paste, drag, upload, camera capture
- Temporary image display
- Reverse image search using Google Lens
- Sends data URLs to backend via Chrome messaging
- Receives garment and selfie images
- Sends both to IDM-VTON on Hugging Face
- Returns generated virtual try-on image
- Hosts temporary garment images for Google Lens search
- Returns a mock preview when free-tier HF quota is exhausted
The extension relies on the Hugging Face ZeroGPU free tier.
Important notes:
- Daily GPU minutes are limited
- IDM-VTON may still treat API calls as anonymous even with a token
- When quota is exceeded:
- Backend returns a mock result with token notice
- Quota resets every 24 hours
- Best workaround: fork IDM-VTON into your own HF account
The extension supports garment reverse-search via Google Lens.
Notes:
- Implemented using "lens.google.com/uploadbyurl?url=
"
- Feature works as of version 1.0 (2025-11)
- Google may change Lens APIs at any time
- Uploaded images are temporarily hosted under /public/reverse/ on the backend
TryMe uses a minimal Node/Express backend built specifically to:
- Accept garment + selfie images from the extension
- Send them to IDM-VTON using the official
@gradio/client - Normalize the Space output
- Return a browser-ready image to the extension
The backend is deployed on Render (free tier).
This means:
- If the service has been idle, it may take 20–40 seconds to spin up
- During cold starts, the extension may temporarily fall back to the mock image
Once warm, results are generally fast.
All virtual try-on results are generated using the excellent open-source model licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License:
🔗 https://huggingface.co/spaces/yisol/IDM-VTON
🔗 https://github.com/yisol/IDM-VTON
This extension does not modify or distribute model weights.
All inference calls go directly through the publicly available Hugging Face Space API using the @gradio/client library.
If you use or extend this project, please credit yisol et al. for their work.