ML: Image classification via fine-tuning with EfficientNet

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102

# https://keras.io/examples/vision/image_classification_efficientnet_fine_tuning/ import numpy as np import tensorflow_datasets as tfds import tensorflow as tf import matplotlib.pyplot as plt import keras from keras import layers from keras.applications import EfficientNetB0 IMG_SIZE=224 # IMG_SIZE is determined by EfficientNet Model choice BATCH_SIZE=64 # Loading data dataset_name="stanford_dogs" (ds_train, ds_test), ds_info=tfds.load( dataset_name, split=["train", "test"], with_info=True, as_supervised=True ) NUM_CLASSES=ds_info.features["label"].num_classes # When the dataset include images with various size, we need to resize them into a shared size. The Stanford Dogs dataset includes only images at least 200x200 pixels in size. Here we resize the images to the input size needed for EfficientNet. size=(IMG_SIZE, IMG_SIZE) ds_train=ds_train.map(lambda image, label:(tf.image.resize(image, size), label)) ds_test=ds_test.map(lambda image, label:(tf.image.resize(image, size), label)) def format_label(label): string_label=label_info.int2str(label) return string_label.split("-")[1] label_info=ds_info.features["label"] for i, (image, label) in enumerate(ds_train.take(9)): ax=plt.subplot(3, 3, i+1) plt.imshow(image.numpy().astype("uint8")) plt.title("{}".format(format_label(label))) plt.axis('off') % Data augmentation img_augmentation_layers=[ layers.RandomRotation(factor=0.15), layers.RandomTranslation(height_factor=0.1, width_factor=0.1), # A preprocessing layer which randomly translates images during training. layers.RandomFlip(), # A preprocessing layer which randomly flips images during training. layers.RandomContrast(factor=0.1), # A preprocessing layer which randomly adjusts contrast during training. ] def img_augmentation(images): for layer in img_augmentation_layers: images=layer(images) return images # Here we plot 9 examples of augmentation result of a given figure. for image, label in ds_train.take(1): for i in range(9): ax=plt.subplot(3, 3, i+1) aug_img=img_augmentation(np.expand_dims(image.numpy(), axis=0)) aug_img=np.array(aug_img) plt.imshow(aug_img[0].astype("uint8")) plt.title("{}".format(format_label(label))) plt.axis('off') def input_preprocess_train(image, label): # One-hot/categorical encoding image=img_augmentation(image) label=tf.one_hot(label, NUM_CLASSES) # Returns a one-hot tensor. return image, label def input_preprocess_test(image, label): label=tf.one_hot(label, NUM_CLASSES) # Returns a one-hot tensor. return image, label ds_train=ds_train.map(input_preprocess_train, num_parallel_calls=tf.data.AUTOTUNE) ds_train=ds_train.batch(batch_size=BATCH_SIZE, drop_remainder=True) ds_train=ds_train.prefetch(tf.data.AUTOTUNE) ds_test=ds_test.map(input_preprocess_test, num_parallel_calls=tf.data.AUTOTUNE) ds_test=ds_test.batch(batch_size=BATCH_SIZE, drop_remainder=True) # Training a model from scratch model=EfficientNetB0( include_top=True, weights=None, classes=NUM_CLASSES, input_shape=(IMG_SIZE, IMG_SIZE, 3), ) model.compile(optimizer='adam', loss="categorical_crossentropy", metrics=["accuracy"]) model.summary() # all model.summary() to print a useful summary of the model, which includes: # Name and type of all layers in the model. # Output shape for each layer. # Number of weight parameters of each layer. # If the model has general topology (discussed below), the inputs each layer receives # The total number of trainable and non-trainable parameters of the model. epochs=20 hist=model.fit(ds_train, epochs=epochs, validation_data=ds_test)