Deep Learning for Computer Vision

Defining placeholders for input data and targets

A placeholder is a tensor where the data is passed. Placeholders aren't specific values but will receive input during computation. The input size of the perceptron, number of classes, batch size, and the total number of iterations or batches are first declared.  x_input is the input where the images will be fed later. y_input is the placeholder where the one-shot labels or targets will be supplied as shown here:

input_size = 784no_classes = 10batch_size = 100total_batches = 200x_input = tf.placeholder(tf.float32, shape=[None, input_size])y_input = tf.placeholder(tf.float32, shape=[None, no_classes])

The None in the shape argument indicates that it can be of any size as we have not yet defined ...