Hand-Written-Digit-Recognition-using_CNN MODEL-(Python)

By Kunjesh Mahajan

In this project, I use the CNN method to recognize the handwritten digits provided by the user.

Convolutional Neural Network  - It is a class of deep neural networks. 

Below is the procedure to make the mnist model.

MNIST handwritten digit classification problem is a standard dataset used in computer vision and deep learning.

train - test - split

(X_train, y_train), (X_test, y_test) = dataset.load_data()

normalize value to b/w 0and1

X_train= X_train/255.0
X_test= X_test/255.0

CNN (BATCH, HEIGHT, WIDTH, 1)

ANN (BATCH_SIZE, FEATURES)

FEATURES = WIDTH * HEIGHT

reshape array to fit in the network.

X_train = X_train.reshape(X_train.shape[0], -1)
X_test = X_test.reshape(X_test.shape[0], -1)

(batch_size, height, width, 1)

ANN

from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Dropout

0-1

model = Sequential()
model.add(Dense(128, activation='relu'))
model.add(Dropout(0.2))

model.add(Dense(128, activation='relu'))
model.add(Dropout(0.2))

0-9

model.add(Dense(10, activation='softmax'))

model.compile('adam', 'sparse_categorical_crossentropy', metrics=['acc'])

model.fit(X_train, y_train, epochs=3, batch_size=12, validation_split=0.1)

making prediction

plt.imshow(X_test[1255].reshape(28,28), cmap='gray')
plt.xlabel(y_test[1255])
plt.ylabel(np.argmax(model.predict(X_test)[1255]))


model.save('digit_trained.h5')

The model is ready.

prediction via paints.