Virtual Try-On FashionGenAI is an AI-powered tool that allows users to visualize themselves in different clothes based on their own images and text prompts. The project utilizes Stable Diffusion Inpainting and the U2Net segmentation model to isolate clothing parts (upper or lower body) and generate a customized outfit based on the user's choice.
- Use pre-trained Stable Diffusion for realistic clothing inpainting.
- Automatic clothing segmentation with U2Net.
- Customizable prompts for clothing style and color.
- Optional background removal using rembg.
-
Clone the repository:
git clone https://github.com/Cyanex1702/Virtual_Try_on_FashionGenAi.git cd Virtual_Try_on_FashionGenAi -
Install dependencies:
pip install -r requirements.txt
-
Download the pre-trained U2Net model:
# Assumes the checkpoint path is already configured # Alternatively, manually place the checkpoint in the `trained_checkpoint` folder
To run the virtual try-on pipeline, use the following command:
python main.py --image <path_to_image> --prompt "A pink cloth" --part upper --resolution 512 --output output.jpg--image: Path to the input image.--prompt: Text prompt describing the desired clothing style.--part: Which body part to focus on (upperorlower).--resolution: Resolution for the generated output.--num_steps: Number of diffusion steps (default 5).--guidance_scale: Strength of prompt adherence (default 7.5).--rembg: Enable background removal.--output: Path to save the final output image.
- Stable Diffusion Inpainting: Utilizes a model trained by RunwayML to fill in the clothing region.
- U2Net Segmentation: Used for precise segmentation of clothing regions (upper or lower body) from images.
- rembg: Background removal for a cleaner result.
Example command:
python main.py --image ./example.jpg --prompt "A blue dress" --part upper --output result.jpgOutput:
- Support for additional clothing items.
- Enhanced fine-tuning for more diverse prompts.
This project leverages the open-source Stable Diffusion and U2Net.