Corpus for Angry Birds Level Generation
Adeel Zafar, Shahbaz Hassan, Qazi Shuja uddin · 2019 2nd International Conference on Computing, Mathematics and Engineering Technologies (iCoMET) · 2019
Levels testify as an important game element for Procedural Content Generation. Recently, machine learning techniques have been used for the automatic content generation of games. However, there is a need for a corpus for training purposes. Considering this, we provide a corpus for the famous angry birds game. Our corpus contains 200 levels in total, with 100 levels from the original Angry Birds game and 100 from the Baseline level generator. These levels are encoded using a tile-based approach and are freely available for applying machine learning techniques. Further to this, our next step would be to mimic the style of these levels in the forthcoming angry birds level generation competition.