The ability of computers to recognize numbers written by humans is called handwritten digit recognition. This is one of the practically significant problems in pattern recognition applications. Applications for digit recognition include form data entry, bank check processing, mail sorting, and more. Since handwritten digits are imperfect and can be generated with a variety of flavors, it is a difficult job for the machine. The answer to this problem is handwritten digit recognition, which uses an image of a digit to identify the digit contained in the image. Using a deep neural network called CNN, we will develop handwritten digit recognition in this project by using the MNIST dataset that contains photos of handwritten digits from zero to nine (Convolutional Neural Network). The handwritten features and much prior knowledge form the basis of traditional handwriting recognition systems. An optical character recognition (OCR) system must be trained using these prerequisites, which is a difficult process. Deep-learning approaches have been the focus of handwriting recognition research in recent years, leading to breakthrough results. An example of a deep-learning algorithm is convolutional neural networks, which use filters to learn the different features of an image as they are fed into the algorithm. Finally, a graphical user interface is created to draw the figure.