slava-qw / face-swap-tg-bot

Face Swapping Microservice

Face Swap Telegram Bot

Overview

This is a simple pet project: a Telegram bot that allows users to seamlessly swap faces in their images. There are two types of swapping: using classical computer vision algorithms (like face meshing) based on facial landmarks obtained from the neural network, and the more accurate method using a pretrained model to swap faces from two given photos. Additionally, it can be deployed using Docker.

Features

• This project uses two types of swapping (all model results can be found either in the original repositories or in the /research/ folder).
  • Face parsing (by OpenCV, Facer, and FaRL)
  • SimSwap model
  • The e4s model showed unsatisfactory results.
• Telegram integration for easy accessibility.
• User-friendly commands with inline image processing.

Getting Started

Follow these steps to set up the bot.

Prerequisites

• Telegram API token (get yours at Telegram BotFather)

Installation

1. Clone the repository:

   git clone https://github.com/slava-qw/face-swap-tg-bot.git
   cd face-swap-tg-bot

2. Write down your Telegram API token in the configuration file (.env). Also, for correct work, get the https-link to set up the webhook (e.g., https://localhost.run/docs/).

Local

2. Install dependencies:

   pip install -r requirements.txt

3. Run the code:

   python main.py

Docker

2. Build a Docker image and run a container:

   docker-compose up