DIGIT RECOGNITION

MOTIVATION:

• The handwritten digit recognition can play a huge role in increasing the speed of the postal systems.
• The letters can be segregated based on the PIN code using this model.

TASK:

• In this project we classify handwritten digits using deep learning.

DATASET:

• The data used here is taken from Tensorflow datasets. To access the dataset click here.

TRAIN and TEST DATA:

DATA	SAMPLE SIZE
Train_data	60,000
Test_data	10,000

IMAGE PIXELS:

• The image in this dataset are grayscale images. The dimension of each image is (28,28,3).
• The height and width of the image is equal to 28.
• Number of channels in the image is 1.

NEURAL_NETWORK ARCHITECTURE:

• In this project we made use of the convolutional neural network.
• The convolutional network architecture is LENET-5.
• LENET-5 is one of the best architecture used for grayscale images.
• 

OUTCOME:

• In this project we are able to predict the handwritten digits with an accuracy of 99%.
•