Building an optical character recognizer using neural networks

Now that we know how to interact with the data, let's build a neural network-based optical character-recognition system.

How to do it…

  1. Create a new Python file, and import the following packages:
    import numpy as np
import neurolab as nl
  2. Define the input filename:
    # Input file
input_file = 'letter.data'
  3. When we work with neural networks that deal with large amounts of data, it takes a lot of time to train. To demonstrate how to build this system, we will take only 20 datapoints:
    # Number of datapoints to load from the input file
num_datapoints = 20
  4. If you look at the data, you will see that there are seven distinct characters in the first 20 lines. Let's define them:
    # Distinct characters
orig_labels = 'omandig'

# Number of distinct characters
num_output = len(orig_labels)
  5. We will use 90% of the data for training and remaining 10% for testing. Define the training and testing parameters:
    # Training and testing parameters
num_train = int(0.9 * num_datapoints)
num_test = num_datapoints - num_train
  6. The starting and ending indices in each line of the dataset file:
    # Define dataset extraction parameters 
start_index = 6
end_index = -1
  7. Create the dataset:
    # Creating the dataset
data = []
labels = []
with open(input_file, 'r') as f:
    for line in f.readlines():
        # Split the line tabwise
        list_vals = line.split('	')
  8. Add an error check to see whether the characters are in our list of labels:
            # If the label is not in our ground truth labels, skip it
        if list_vals[1] not in orig_labels:
            continue
  9. Extract the label, and append it the main list:
            # Extract the label and append it to the main list
        label = np.zeros((num_output, 1))
        label[orig_labels.index(list_vals[1])] = 1
        labels.append(label)
  10. Extract the character, and append it to the main list:
            # Extract the character vector and append it to the main list
        cur_char = np.array([float(x) for x in list_vals[start_index:end_index]])
        data.append(cur_char)
  11. Exit the loop once we have enough data:
            # Exit the loop once the required dataset has been loaded
        if len(data) >= num_datapoints:
            break
  12. Convert this data into NumPy arrays:
    # Convert data and labels to numpy arrays
data = np.asfarray(data)
labels = np.array(labels).reshape(num_datapoints, num_output)
  13. Extract the number of dimensions in our data:
    # Extract number of dimensions
num_dims = len(data[0])
  14. Train the neural network until 10,000 epochs:
    # Create and train neural network
net = nl.net.newff([[0, 1] for _ in range(len(data[0]))], [128, 16, num_output])
net.trainf = nl.train.train_gd
error = net.train(data[:num_train,:], labels[:num_train,:], epochs=10000, 
        show=100, goal=0.01)
  15. Predict the output for test inputs:
    # Predict the output for test inputs 
predicted_output = net.sim(data[num_train:, :])
print "
Testing on unknown data:"
for i in range(num_test):
    print "
Original:", orig_labels[np.argmax(labels[i])]
    print "Predicted:", orig_labels[np.argmax(predicted_output[i])]
  16. The full code is in the ocr.py file that's already provided to you. If you run this code, you will see the following on your Terminal at the end of training:
    How to do it…

    The output of the neural network is shown in the following screenshot:

    How to do it…
..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset
3.145.44.174