The problem goal is to classify an input image into one of 9 classes. The input drawing represents hand written shapes. This recognition is useful in applications that takes hand drawn shapes from the user for example through a Tablet and tries to recognize them. Applications could be recognizing hand drawn electronic circuit or UML diagram.
This dataset is manually collected from a group of people asked to draw those shapes, then those hand written shapes saved as bmp images to the computer. The shapes we have are Circle, line, triangle, ellipse, ... Images are 256x256 pixel. Dataset contains around 250 training example for each of the 9 classes and around 120 image for each class for testing.