PixelizationBot

PixelizationBot is a Telegram bot that allows you to pixelate images. It uses machine learning models to process images and provides a convenient interface for interaction via Telegram.

Demo:

@pixelizationaibot

Installation

1. Clone the repository:

   git clone https://github.com/yourusername/PixelizationBot.git
   cd PixelizationBot

2. Install the required dependencies:

   python3 -m venv venv
   source venv/bin/activate
   
   pip install -r requirements.txt

3. Download the models:

   Our software download models automatically from HuggingFace 🤗

   But you can this yourself: Download all three models from the table and place them into the /models directory inside the extension.

   URL	Filename
   Model 1 Link	pixelart_vgg19.pth
   Model 2 Link	alias_net.pth
   Model 3 Link	160_net_G_A.pth

   Alternative download link for models:

   • Pixeldrain
   • HuggingFace 🤗

Configuration Setup

1. Create a configuration file:

Copy config/config.yml.example to config/config.yml and edit it according to your needs.

cp config/config.yml.example config/config.yml

2. Configuration parameters:

   • Bot config:
     • API_ID and API_HASH: Your Telegram application identifiers.
     • API_TOKEN: Your bot's token.
   • Model config:
     • NETG_PATH, ALIASNET_PATH, VGG19_PATH: path where place models
     • REFERENCE_PATH: path where stored coloro scheme image reference
   • Worker config:
     • FORCE_USE_CPU: Force to use CPU (if you have cuda card in your computer)
     • NUM_PROCESS: Number of workers for image processing.
     • NUM_TORCH_THREADS: How many CPU cores will be used.
     • MODEL_KEEP_ALIVE_SECONDS: Time in seconds for which models remain loaded during inactivity.
   • Queue config:
     • SLOTS_QUANTITY: How many one user can send pictures in queue.
     • DELAY_STATUS: How long before we need to update the queue message in seconds

   Also you can use flag --config when start script, and set custom path for your configuration

   python3 main.py # looking in work folder by path config/config.yml
   
   python3 main.py --config /home/foxgirl/config/config.yml # take config by custom path

Example systemd Service

[Unit]
Description=Pixelization ML model Telegram bot.

[Service]
Restart=on-abort
Type=simple
ExecStart=PixelizationBot/venv/bin/python3 main.py # Set path to 
WorkingDirectory=PixelizationBot #Path where installed PixelizationBot

[Install]
WantedBy=default.target

Credits

• txlyre - prototype demo
• Th3ro - memory leak fix and support with another developing
• Taruu - Idea author and developing

Original repo project:

• https://github.com/WuZongWei6/Pixelization

TODO

• Make better queue system (always can be better)
• Remade Model network from scrap
• make custom lib for better conversation