/Clash-Royale-AI-Card-Tracker

An application that trains 2 convolutional neural networks to be able to make predictions during a live match (real-time application), on the opponent's card deck, card cycle, and current elixir.

Primary LanguagePythonApache License 2.0Apache-2.0

Clash Royale AI Card Tracker

Currently, the code is not commented very well. This will change soon.

I wrote a report explaining how this AI Assistant was made. It's tailored towards an non-computer-science audience. Please see: Rough Draft Report

Libraries Used:

  • openCV (Image Preprocessing)
  • Keras/TensorFlow (Convolutional Neural Networks)
  • TkInter (GUI)
  • PIL (Mapping images into GUI)

If you still want to see the code for the following:

  1. The Convolutional Neural Network Architecture (LeNet).

  2. Training the first CNN (Convolutional Nerual Network).

  3. Predictions on the first CNN.

  4. Live (Real-Time) predictions test on the first CNN, to learn opponent's deck.

  5. Training the second CNN (Convolutional Nerual Network).

  6. Predictions on the second CNN.

  7. Live (Real-Time) predictions test on the second CNN, to know which card my opponent is playing.

  8. Both CNN's running together during real-time to predict opponent's card cycle + elixir. (This section also has a lot of code regarding the actual AI application).

  9. TkInter GUI

The code can be found here: Clash_Royale_Helper.py

AI Assistant