Practicing how to fine-tune NN models designed for image segmentation tasks, beware lots of comments in code (trying to learn!).

Follow these steps to run this project on your local machine:

1. Clone this repository:

   git clone https://github.com/your-username/clothing-segmentation.git
   cd clothing-segmentation
   

2. Download the Clothing-Co-Parsing dataset from bearpaw's repository.

3. Set up the directory structure:

   mkdir -p data/annotations data/input_images/images
   

4. Add the annotations from the dataset:

   • Place the image-level folder in data/annotations/
   • Place the pixel-level folder in data/annotations/

5. Add the images from the dataset:

   • Place all the images in data/input_images/images

6. Run the pre-processing script:

   python -m scripts.pre_processing
   

   This will create masks in data/input_images/masks/ and distribute images into train and validation dataset folders.

7. Start the training process:

   python -m scripts.train
   

Before installing the dependencies, it's recommended to create a virtual environment:

1. Create a virtual environment:

   python -m venv venv
   

2. Activate the virtual environment.

3. Install the required dependencies using the requirements.txt file:

   pip install -r requirements.txt
   

Dataset from: https://github.com/bearpaw/clothing-co-parsing

@inproceedings{yang2014clothing,
  title={Clothing Co-Parsing by Joint Image Segmentation and Labeling},
  author={Yang, Wei and Luo, Ping and Lin, Liang}
  booktitle={Computer Vision and Pattern Recognition (CVPR), 2014 IEEE Conference on},
  year={2013},
  organization={IEEE}
}

References code from: https://github.com/IzPerfect/Clothing-Segmentation/tree/master