TeleAI

You can test the bot by setting it up yourself.

The model has been trained, and the bot is ready to go. However, if you wish to see the training in action, you may run the training.py file after installing all the necessary packages.

About this bot

This bot has been trained using OpenCV, Tensorflow and the keras deep learning API, using the CIFAR10 dataset. Since this is purely a demo project, the smaller and simpler CIFAR10 was chosen over CIFAR100 or other more advanced datasets.

It was built using the Python programming language.

Capabilities

The teleAI bot is a telegram bot capable of image recognition, and can recognise images from the following categories:

Plane, Car, Bird, Cat, Deer, Dog, Frog, Horse, Ship, Truck

Set up - If you wish to test it out

Make sure you have Python3 installed. I am using Python 3.11.2

Clone the repository:

  1. Navigate to an empty directory.
  2. In a git bash terminal, type git clone https://github.com/oeggy03/TeleAI.git

Install the packages by typing:

pip install -r ./requirements.txt

Setting up a telegram bot

  1. On your telegram app, go to https://t.me/BotFather
  2. Type and send /newbot.
  3. Follow the instructions, and you will get an API key.
  4. Create a .env file, and type KEY = <Your API key here>
  5. Also take note of the bot’s telegram link (looks like t.me/<bot_username>)

Testing out the bot

  1. Run the main.py file using python main.py in the terminal.
  2. Go to the bot’s telegram link and type in /start.
  3. Upload an image of one of the objects in the mentioned categories.
  4. Done! The bot should be able to recognise the image.

Please note that AI is not perfect, and will classify objects wrongly at times, especially if the object is dark / shadowy. However, it should work well as long as the subject of the photo is obvious.