The dataset contains 16,000 images of four shapes: square, star, circle, and triangle.Each image is 200×200 pixels.

The data was collected using a Garmin Virb 1080p action camera. The shapes were cut from poster board and then painted green. The data processor held each shape in front of the camera for two minutes, moving and rotating the shape as the camera recorded it. The four videos were then processed using OpenCV in Python. The green shape was cropped from the image using color space and resized to 200×200 pixels. The data is divided into four folders: square, circle, triangle, and star. The images are labeled 0.png, 1.png, etc…