2D Geometric shapes with properties

Published: 1 May 2024| Version 1 | DOI: 10.17632/d35tjjgjnx.1


The presented data is a set of 3,80,000 black and white images of a fixed resolution of 200 x 200 pixels Each image consists of one of the geometric shapes: Circle, Triangle, square, Rectangle, Rhombus, Parallelogram, Trapezium with one of the properties: diagonal, perpendicular, chord, radius, diameter., symbol showing sides are equal. In total 76 such classes are created. The background color is white and the foreground color when the shape is drawn is black. o A random rotation angle between -180° and 180°. A random shape perimeter. A random shape position inside the containing image. This dataset can be used for identification of shape using classification approach with deep learning. Choosing a 2D geometrical form dataset also helps understand the number of patterns in each class of particular shape.



Visvesvaraya National Institute of Technology


Geometry, Image Classification, Pattern Recognition, Deep Learning