Add_assignment_3

assignment_3/01_Res50_Predict.py

@@ -0,0 +1,31 @@
+# TensorFlow and tf.keras
+import tensorflow as tf
+from tensorflow.keras.applications.imagenet_utils import decode_predictions
+# Helper libraries
+import numpy as np 
+import cv2
+
+# load an image
+Testing_img_path ="D:\ISE\Internet of Things\RPi_Lab 3\cat2.jpg"
+input_image = cv2.imread(Testing_img_path)
+
+# Preprae data
+input_image = cv2.resize(input_image, (224, 224))
+input_image = np.expand_dims(input_image, axis=0) # Change the shape of image array like input image when trained
+
+# Load the ResNet50 model
+loadpath = "Resnet50.h5"
+resnet_model = tf.keras.models.load_model(loadpath)
+
+# get the predicted probabilities for each class
+predictions = resnet_model.predict(input_image)
+print("Predicted Top 3 is:", decode_predictions(predictions, top=3)[0])
+Answer = decode_predictions(predictions, top=1)[0][0][1]
+print("Predicted = ", Answer)
+
+
+# Visualize, check the answer
+img = cv2.imread(Testing_img_path)
+
+cv2.waitKey(0)
+cv2.destroyAllWindows()

assignment_3/01_Res50_export.py

@@ -0,0 +1,10 @@
+# TensorFlow and tf.keras
+import tensorflow as tf
+from keras.applications import resnet50
+
+# Load the ResNet50 model
+resnet50_model = resnet50.ResNet50(weights='imagenet')
+
+# Save model and Export
+savepath = "Resnet50.h5"
+resnet50_model.save(savepath)

assignment_3/02_detect_number_train.py

@@ -0,0 +1,30 @@
+# TensorFlow and tf.keras
+import tensorflow as tf
+from tensorflow import keras
+
+#load data set from tf.keras.datasets.mnist (Train number and Test number)
+mnist = tf.keras.datasets.mnist # 28x28 image of hand-written digits 0-9
+(x_train, y_train),(x_test, y_test) = mnist.load_data()
+
+# Build Model
+model = keras.Sequential([
+    keras.layers.Flatten(input_shape=(28, 28)),    # takes our 28x28 and makes it 1x784
+    keras.layers.Dense(128, activation='relu'),    # a simple fully-connected layer, 128 units, relu activation
+    keras.layers.Dense(128, activation='relu'),    # a simple fully-connected layer, 128 units, relu activation
+    keras.layers.Dense(10, activation='softmax')   # our output layer. 10 units for 10 classes. Softmax for probability distribution
+])
+# Compile model
+model.compile(optimizer='adam',
+              loss='sparse_categorical_crossentropy',
+              metrics=['accuracy'])
+# Train the model
+model.fit(x_train, y_train, epochs=5)
+
+# Evaluate accuracy
+test_loss, test_acc = model.evaluate(x_test,  y_test)
+print("Test loss:", test_loss)
+print("Test accuracy:", test_acc)
+
+# ***Save model***
+savepath = "num_reader.h5"
+model.save(savepath)

assignment_3/09_detect_number_predict.py

@@ -0,0 +1,30 @@
+# TensorFlow and tf.keras
+import tensorflow as tf
+from tensorflow import keras
+
+# Helper libraries
+import numpy as np
+import cv2
+
+# load an image
+Testing_img_path = "./Testing_images/0.jpg"
+input_image = cv2.imread(Testing_img_path,cv2.IMREAD_GRAYSCALE)
+
+# Preprae data
+input_image = cv2.resize(input_image,(28, 28))
+input_image = cv2.bitwise_not(input_image) # invert Black to White like data input when trained (Line is White)
+input_image = np.expand_dims(input_image, axis=0) # Change the shape of image array like input image when trained
+
+# ***load model***
+loadpath = "num_reader.h5"
+model = tf.keras.models.load_model(loadpath)
+
+# Make predictions
+predictions = model.predict(input_image)
+print(np.argmax(predictions))
+
+# Visualize, check the answer
+Original_input_image_testing = cv2.imread(Testing_img_path)
+cv2.imshow("This is what your computer has seen.",Original_input_image_testing)
+cv2.waitKey(0)
+cv2.destroyAllWindows()

assignment_3/tf_test.py

@@ -0,0 +1,5 @@
+import tensorflow as tf
+
+dataset1 = tf.data.Dataset.from_tensor_slices(tf.random.uniform([4, 10]))
+
+print(dataset1.element_spec)