Multimodal Dataset for Connect Four

Dataset contains six main components derived from human-versus-human Connect Four games played on a physical board, as well as computer-generated optimal move data. Dataset Components Video Dataset – Recordings of 52 human-versus-human Connect Four matches on a physical board. Board Annotations – Frame-wise annotated boards with grid coordinates. Piece Annotations – Annotated discs for both players and empty grids Image Vectors – Vectorized board representations derived from frames. Optimal Move Dataset – One million unique board positions with optimal moves computed using a minimax-based solver. Annotation Tool – A Python-based GUI application for frame-to-symbolic board conversion and annotation consistency checking.