Cheater's Inn

This is my submission for the AI & Game Jam 2021.

Play a card game where each AI opponent is a neural network that was trained to play by a different set of rules. Whenever you have the highest or the lowest hand, you can accuse your fellow players of cheating. Opponents that are correctly identified can be defeated and someone else will take their place. Unmask enough opponents to reach the boss battle!

Post-jam changes:

- fixed dependency issue causing crashes on certain devices;

- fixed two bugs in the training framework and retrained all neural networks;

- added sound effects;

- capped framerate to 60;

- slightly changed the boss battle.

This game and the framework used to train the neural networks are open source. Their source code can be found under "More information".

The following free assets were used:

- "2-bit Doodle People" by Buch, with modified color palette;

- Playing cards by Nora Shishi, rh0, devurandom and Kumpu, with modified color palette;

- 1-bit Pack, casino sounds and RPG sounds by Kenney;

- Texturina font by Guillermo Torres, Omnibus-Type.