ML: Training an image classifier from scratch on the Kaggle Cats vs Dogs dataset

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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142

# https://keras.io/examples/vision/image_classification_from_scratch/ import os import numpy as np import keras from keras import layers from tensorflow import data as tf_data import matplotlib.pyplot as plt num_skipped=0; # the number of photos deleted for folder_name in ("Cat", "Dog"): folder_path=os.path.join("D:\PyTest\PetImages", folder_name) for fname in os.listdir(folder_path): fpath=os.path.join(folder_path, fname) try: fobj=open(fpath, "rb") # "r" - Read & "b" - Binary - Binary mode is_jfif=b"JFIF" in fobj.peek(10) finally: fobj.close() if not is_jfif: num_skipped= num_skipped+1 os.remove(fpath) print(f"Deleted {num_skipped} images.") image_size=(180, 180) # width, height batch_size=128 train_ds, val_ds=keras.utils.image_dataset_from_directory( "D:\PyTest\PetImages", validation_split=0.2, # fraction of data to reserve for validation subset="both", seed=1337, # Optional random seed for shuffling and transformations. image_size=image_size, batch_size=batch_size, ) plt.figure(figsize=(10, 10)) # plot the images for images, labels in train_ds.take(1): for i in range(9): ax=plt.subplot(3, 3, i+1) plt.imshow(np.array(images[i]).astype("uint8")) # plt.show() plt.title(int(labels[i])) plt.axis("off") data_augmentation_layers=[ layers.RandomFlip("horizontal"), # which flip mode to use. Can be "horizontal", "vertical", or "horizontal_and_vertical". "horizontal" is a left-right flip and "vertical" is a top-bottom flip. Defaults to "horizontal_and_vertical" layers.RandomRotation(0.1), # a float represented as fraction of 2 Pi, or a tuple of size 2 representing lower and upper bound for rotating clockwise and counter-clockwise. ] def data_augmentation(images): # a function is defined, all input images will be augmented and return. for layer in data_augmentation_layers: images=layer(images) return images plt.figure(figsize=(10, 10)) # plot the images for images, _ in train_ds.take(1): for i in range(9): augmented_images=data_augmentation(images) ax=plt.subplot(3, 3, i+1) plt.imshow(np.array(augmented_images[0]).astype("uint8")) plt.axis("off") inputs=keras.Input(shape=image_size) x=data_augmentation(inputs) # Call a function, we defined above. x=layers.Rescaling(1./255)(x) # To rescale an input in the [0, 255] range to be in the [0, 1] range, you would pass scale=1./255. # Apply 'data_augmentation' to the training images. train_ds=train_ds.map( lambda img, label: (data_augmentation(img), label), num_parallel_calls=tf_data.AUTOTUNE, ) # Prefetching samples in GPU memory helps maximize GPU utilization. train_ds=train_ds.prefetch(tf_data.AUTOTUNE) val_ds=val_ds.prefetch(tf_data.AUTOTUNE) def make_model(input_shape, num_classes): # Define a function inputs=keras.Input(shape=input_shape) #Entry block x=layers.Rescaling(1.0/255)(inputs) # To rescale an input in the [0, 255] range to be in the [0, 1] range, you would pass scale=1./255. x=layers.Conv2D(128,3,strides=2, padding="same")(x) # int, the dimension of the output space (the number of filters in the convolution). x=layers.BatchNormalization()(x) # Layer that normalizes its inputs. x=layers.Activation("relu")(x) previous_block_activation=x for size in [256, 512, 728]: x=layers.Activation("relu")(x) x=layers.SeparableConv2D(size, 3, padding="same")(x) # Depthwise separable 2D convolution x=layers.BatchNormalization()(x) # Layer that normalizes its inputs. x=layers.Activation("relu")(x) x=layers.SeparableConv2D(size, 3,padding="same")(x) x=layers.BatchNormalization()(x) x=layers.MaxPooling2D(3, strides=2, padding="same")(x) # Max pooling operation for 2D spatial data. residual=layers.Conv2D(size, 1, strides=2, padding="same")( previous_block_activation ) x=layers.add([x,residual]) previous_block_activation=x x=layers.SeparableConv2D(1024, 3, padding="same")(x) x=layers.BatchNormalization()(x) x=layers.Activation("relu")(x) x=layers.GlobalAveragePooling2D()(x) if num_classes==2: units=1 else: units=num_classes x=layers.Dropout(0.25)(x) # We specify activation=None so as to return logits outputs=layers.Dense(units, activation=None)(x) return keras.Model(inputs, outputs) img=keras.utils.load_img("D:\weed\PyTest\PetImages\Cat\6779.jpg", target_size=image_size) img_array=keras.utils.img_to_array(img) img_array=keras.ops.expand_dims(img_array, 0) predictions=model.predict(img_array) score=float(keras.ops.sigmoid(predictions[0][0])) print(f"This image is {100*(1-score):.2f}% cat and {100*score:.2f}% dog.")