Telegram bot for image enhancement

This paper uses:

• Convolutional Neural Network (CNN)
• Residual Dense Block (RDB)
• Telegram bot as a user interface
• ReLU activation function
• ImageNet dataset taken from Kaggle
• Image normalization
• Convert images from RDB to YUV
• Numerical metrics: PSNR, SSIM, MSE, MAE
• Visual comparison of results

Cloning a repository

To download this code from GitHub, open a terminal and run the following commands:

git clone https://github.com/TokenRR/Bachelor_thesis_on_image_enhancement.git
cd Bachelor_thesis_on_image_enhancement

Installing dependencies

Create a virtual environment and install dependencies:

python -m venv venv
source venv/bin/activate  # For Windows, use venv\Scripts\activate
pip install -r requirements.txt

Setting up the configuration

Open the code/bot/config.py file and change the bot token to your own:

TOKEN='your-token-here'

Start Telegram bot

To start the Telegram bot, go to the code/bot directory and run the following command:

python main.py

Training models

There are already trained models in the code\models directory, but if you want to make changes, you can edit the code in the corresponding Jupyter Notebooks files.