How to integrate own images into this machine learning code by Dr. Martin?

I am fascinated by the code for machine learning by Dr. Martin at Github. This code gives very good accuracy with mNist data set. Now I would like to use this code to do some ML work with my own color images for a different application. I am new to ML but also to Python and learning. I got help to find where the mNist data is fed to the network.

Loading the images at line 37 : # Download images and labels into mnist.test (10K images+labels) and mnist.train (60K images+labels) mnist = mnistdata.read_data_sets(“data”, one_hot=True, reshape=False)

Feeding to the network at Line 98 : ….

# You can call this function in a loop to train the model, 100 images at a time def training_step(i, update_test_data, update_train_data):      # training on batches of 100 images with 100 labels     batch_X, batch_Y = mnist.train.next_batch(100)      # compute training values for visualisation     if update_train_data:         a, c, im, w, b = sess.run([accuracy, cross_entropy, I, allweights, allbiases], {X: batch_X, Y_: batch_Y})         print(str(i) + ": accuracy:" + str(a) + " loss: " + str(c) + " (lr:" + str(learning_rate) + ")")         datavis.append_training_curves_data(i, a, c)         datavis.update_image1(im)         datavis.append_data_histograms(i, w, b) 

However, I still can not figure out how to change this to load and feed my own images to the network.

Found one example at here.

How can I change the existing mNist loading with my own images? I would like to have my own mNist data set created with own images so that I could use with existing code.