Skip to content

Latest commit

 

History

History
107 lines (78 loc) · 2.8 KB

README.md

File metadata and controls

107 lines (78 loc) · 2.8 KB

DIVIDEND 🤑

Means - Dynamic Integration for Video Improvement and Digital Eradication of Non-Desired Data

FUNTIONALITY

Fr ? Read Description for functionality

Features

  • Automated Watermark Detection
  • High-Resolution Output
  • Temporal Consistency
  • Plug-and-Play Integration
  • Scalable and Robust

Project Structure (ofcourse i'd use gpt what do you think am i ? a legacy coder uhh ??)

DIVIDEND/
│
├── datasets/                # Training and testing datasets
│   ├── train/
│   ├── test/
│
├── models/                  # Model architectures
│   ├── unet_attention.py     # U-Net with attention mechanism
│   ├── discriminator.py      # GAN-based discriminator
│   └── temporal_network.py   # Temporal consistency model (3D CNN)
│
├── scripts/                 # Training and utility scripts
│   ├── train.py              # Training script
│   ├── test.py               # Testing script
│   └── utils.py              # Utility functions (data loading, pre-processing)
│
├── checkpoints/             # Saved model weights
├── results/                 # Output video frames
├── requirements.txt         # Python dependencies
└── README.md                # Project documentation

Getting Started

1. Clone the repository:

git clone https://github.com/venusarathy/dividend.git

2. Install dependencies:

pip install -r requirements.txt

3. Prepare your dataset:

  • Place your watermarked and non-watermarked video frames in the datasets/ directory.
  • Ensure the dataset is properly structured into train/ and test/ directories.

4. Train the model:

python scripts/train.py

5. Test the model:

python scripts/test.py

Models

  1. U-Net with Attention (unet_attention.py):

    • Detects and removes watermarks using an attention mechanism to focus on watermark regions.

  2. Temporal Consistency Network (temporal_network.py):

    • Ensures smooth transitions between video frames using 3D convolutions.

  3. Discriminator (discriminator.py):

    • Used in GAN architecture to enhance the realism of generated frames during adversarial training.

Contributions

  • Contributions are welcome! Please submit a pull request or open an issue for any bugs or feature requests.
  • See you on the pull request tab :)

License

This project is licensed under the MIT License. See the LICENSE file for details.

Author

Venu Sarathy
GitHub Profile

Acknowledgments

Special thanks to all contributors and supporters.