Grid's and Mazes Dataset

Dataset Content

• Images: The dataset includes images of grids and mazes with a target aspect ratio of 16:9. Each image has a resolution greater than or equal to 1280×720 pixels, allowing for high-quality analysis.
• Distribution: There is an uneven distribution of data, with 10 maze images for each grid image. Users can duplicate grid images or employ data augmentation techniques to balance the dataset.

Data Collection Methodology

• Generation Process: Images are algorithmically generated to meet the specified resolution and aspect ratio requirements. The dataset includes variations in wall and cell pixel sizes to provide a diverse set of images.
• File Naming Convention: Images are named using the format “{X Width}x{Y Width}{Wall Pixel Size}{Cell Pixel Size}”. Maze images include an additional run counter and an identifier (“g” for grid, “m” for maze) to distinguish them.

Applications

• Classification Models: Train and evaluate models to classify images as either grids or mazes.
• Image Transformation Detection: Utilize the grid generation as a baseline for future image transformation detection datasets and tasks.