Conversational Image Retreival with Virtual Try-On

Requirements

We assume that you are in the repository directory. We highly recommend you to create conda environment by conda create -n cigar python=3.8 ipython

1. Clone our repository

2. Image Retrieval

Our pipeline for image retrieval is folked from https://github.com/nashory/FashionIQChallenge2020.

1. Install all dependencies.

   pip install -r ir_requirements.txt
   

2. Download dataset.

   cd image_retrieval/retrieval_model/
   bash run_download_image.sh

3. Download word embedding for user feedback.

   pip install gdown
   cd image_retrieval/retrieval_model/ours/train
   gdown --fuzzy https://drive.google.com/file/d/1gLl73829eVZrWuXQerJpsz8v9mPQRMAX/view?usp=sharing
   tar -xvf assets.tar.gz

4. Download checkpoint for text encoder in model: ask admin (currently not supported)

5. Download FashionIQ dataset

3. Garment Transfer

Our pipeline for garment transfer is folked from https://github.com/cuiaiyu/dressing-in-order.git

1. Parser

• pip install ninja
• Download checkpoint from here.

2. Pose Transder

• Install dependencies of pose transfer.

cd pose
pip install -r requirements.txt

3. Dressing-in-order(DIOR)

• Install dependencies of dior.

cd dior
pip install -r requirements.txt

• Download checkpoint of pretrained model and unzip at checkpoints/

gdown --fuzzy https://drive.google.com/file/d/1JvLu6RJ4QBAYf6ON9i_DWU3Jlj56vz5P/view?usp=sharing

Run

python main.py