Classification of coil100 images using keras

coil100-deeplearning.py

import os
#Work around for https://github.com/tensorflow/tensorflow/issues/24496
os.environ['TF_FORCE_GPU_ALLOW_GROWTH'] = 'true'

# Work around for https://github.com/tensorflow/tensorflow/issues/33024
import tensorflow.compat as compat
compat.v1.disable_eager_execution()

import imageio
from sklearn.model_selection import KFold, train_test_split
from skimage.color import rgb2gray
from matplotlib import pyplot as plt
import tensorflow.keras as keras
import numpy as np

def GetCoil100FileList(pth):

	objects = []

	for i in range(1, 101):
	
		fileList = []
		for ang in range(0, 360, 5):

			fina = os.path.join(pth, "obj{}__{}.png".format(i, ang))
			if os.path.exists(fina):
				fileList.append(fina)
			else:
				raise RuntimeError("Missing file in coil100: {}".format(fina))
		
		objects.append(fileList)			

	return objects

def LoadCoil100(pth):

	objects = GetCoil100FileList(pth)

	dataX = []
	dataY = []	
	for label, obj in enumerate(objects):
		for fina in obj:
			img = imageio.imread(fina)
			dataX.append(img)
			dataY.append(label)

	return dataX, dataY

def ScaleImgData(dataX):

	dataXout = []

	for img, y in zip(dataX, dataY):
		#img2 = rgb2gray(img)
		img2 = np.array(img, dtype=np.float32) / 255.0
		dataXout.append(img2)

	dataXout = np.array(dataXout)
	return dataXout.reshape(dataXout.shape[0], dataXout.shape[1], dataXout.shape[2], 3)

def GetBasicModel():
	# https://machinelearningmastery.com/how-to-develop-a-convolutional-neural-network-from-scratch-for-mnist-handwritten-digit-classification/

	model = keras.models.Sequential()
	model.add(keras.layers.Conv2D(32, (3, 3), activation='relu', kernel_initializer='he_uniform', input_shape=(128, 128, 3)))
	model.add(keras.layers.MaxPooling2D((2, 2)))
	model.add(keras.layers.Flatten())
	model.add(keras.layers.Dense(200, activation='relu', kernel_initializer='he_uniform'))
	model.add(keras.layers.Dense(100, activation='softmax'))
	# compile model
	opt = keras.optimizers.SGD(lr=0.01, momentum=0.9)
	model.compile(optimizer=opt, loss='categorical_crossentropy', metrics=['accuracy'])
	return model

# plot diagnostic learning curves
def summarize_diagnostics(histories):
	for i in range(len(histories)):
		# plot loss
		plt.subplot(2, 1, 1)
		plt.title('Cross Entropy Loss')
		plt.plot(histories[i].history['loss'], color='blue', label='train')
		plt.plot(histories[i].history['val_loss'], color='orange', label='test')
		# plot accuracy
		plt.subplot(2, 1, 2)
		plt.title('Classification Accuracy')
		plt.plot(histories[i].history['accuracy'], color='blue', label='train')
		plt.plot(histories[i].history['val_accuracy'], color='orange', label='test')
	plt.show()

def RunTests(dataX, dataY):

	X_trainVal, X_test, y_trainVal, y_test = train_test_split(dataX, dataY, test_size=0.25)
	
	X_train, X_val, y_train, y_val = train_test_split(X_trainVal, y_trainVal, test_size=0.25)

	model = GetBasicModel()
	
	history = model.fit(X_train, y_train, epochs=10, batch_size=32, validation_data=(X_val, y_val), verbose=0)

	_, acc = model.evaluate(X_test, y_test, verbose=0)
	print ("Performance", acc)

	summarize_diagnostics([history])

if __name__=="__main__":

	dataX, dataY = LoadCoil100("/home/tim/Downloads/coil-100")

	dataX = ScaleImgData(dataX)
	dataY = keras.utils.to_categorical(dataY)

	RunTests(dataX, dataY)