• PyTorch for Model Training
  
• Streamlit for making the web Application

The following datasets were used to train the model
Base

1. ImageNet 12k

Landscape

1. Landscape Recognition | Image Dataset | 12k Images
2. Landscape Pictures

Person

1. Human Faces
2. FFHQ Face Data Set

Fruits and Flowers

1. Fruits 360

Animals

1. Animals Detection Images Dataset

You can use the pretrained weights avialable under models folder. Or if you want you can train on your own custom dataset

1. Clone the repo https://github.com/rimo10/Image-Colorization-Project.git
2. Open Image_Colorizer.ipynb
3. Copy the model path and run the cell -

def load_checkpoint(model,path):
    model.load_state_dict(torch.load(path)) 
    return model
    
load_checkpoint(model,your_weights_path)

1. Clone the repo git clone https://github.com/rimo10/Image-Colorization-Project.git
2. If you don't have streamlit installed then pip install streamlit
3. Go to the root directory of project and run streamlit run -main.py in the terminal . This will open an interactive web application. Now drag and drop and see the Magic

Paper: Deep Koalarization: Image Colorization using CNNs and Inception-ResNet-v2

All rights reserved under MIT License