Exploring-Tensorflow

Table Of Contents

---

• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 

Useful code snippets

Importing library snippets

import tensorflow as tf
from tensorflow.keras.preprocessing.image import ImageDataGenerator 
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Conv2D, Flatten, Dropout, MaxPooling2D, Activation,GlobalMaxPooling2D,GlobalAveragePooling2D,BatchNormalization
from tensorflow.keras import optimizers
from tensorflow.keras.preprocessing import image #PIL image
from tensorflow.keras.callbacks import EarlyStopping, ReduceLROnPlateau
import tensorflow.keras.backend as keras_backend

import numpy as np
import matplotlib.pyplot as plt
import matplotlib.image as mpimg

Plotting snippet

import matplotlib.pyplot as plt
%matplotlib inline
# list all data in training
print(training.history.keys())
# summarize training for accuracy
plt.plot(training.history['accuracy'])   # training is the variable from the fit method
plt.plot(training.history['val_accuracy'])
plt.title('model accuracy')
plt.ylabel('accuracy')
plt.xlabel('epoch')
plt.legend(['train', 'test'], loc='upper left')
plt.show()
# summarize traning for loss
plt.plot(training.history['loss'])
plt.plot(training.history['val_loss'])
plt.title('model loss')
plt.ylabel('loss')
plt.xlabel('epoch')
plt.legend(['train', 'test'], loc='upper left')
plt.show()

Display image snippet

CLASSES=  {0:"Angry", 1:"Disgust", 2:"Fear", 3:"Happy", 4:"Sad", 5:"Surprise", 6:"Neutral"}

plt.figure(figsize=(12, 12))
for i in range(0,9):
    plt.subplot(5, 3, i+1)
    plt.imshow(image_list[i], cmap="gray") #'image_list' is the list of images
    plt.xlabel(CLASSES[labels[i]]) # 'labels' is the list of labels
plt.tight_layout()
plt.show()

Using pre-trained model snippets

from tensorflow.keras.applications import MobileNetV2  #name of the model to be used
from tensorflow.keras.models import Model #API for the wrapping

# different operations on the pre-trained model
pretrained_model= MobileNetV2(include_top=False,weights='imagenet',input_shape=input_shape)

# List of layers in the pretrained_model
layer_list = [layers.name for layers in pretrained_model.layers]

# make all the layers non-trainable
for layers in pretrained_model.layers:
    layers.trainable = False

# make the layers(by api names) as non-trainable 
for layers in pretrained_model.layers:
     if layers._keras_api_names[0] == 'keras.layers.BatchNormalization':
         layers.trainable = False

#to find the methods available
dir(pretrained_model)

Compiling and fitting model snippets

earlystop = EarlyStopping(patience=10)
learning_rate_reduction = ReduceLROnPlateau(monitor='val_accuracy', 
                                            patience=2, 
                                            verbose=1, 
                                            factor=0.5, 
                                            min_lr=0.00001)

callbacks = [earlystop, learning_rate_reduction]

Mobilenet_v2_model.compile(loss='binary_crossentropy',
              optimizer=optimizers.Adam(learning_rate=0.001),
              metrics=['accuracy'])

training = Mobilenet_v2_model.fit(train_generator,
                   steps_per_epoch=100,epochs=50,
                   validation_data=validation_generator,
                       validation_steps=100,
                       callbacks=callbacks)

Prediction snippets

https://www.tensorflow.org/api_docs/python/tf/keras/preprocessing/

from tensorflow.keras.preprocessing import image
import numpy as np
img_pred = image.load_img("test_set/test_set/dogs/dog.4003.jpg",target_size=(150,150))

img_pred=image.img_to_array(img_pred)
img_pred=np.expand_dims(img_pred, axis=0)

result = Efficientnet_model.predict(img_pred)

Augmentation snippets pre-processing layers for augmentation

from tensorflow.keras.models import Sequential
from tensorflow.keras import layers
img_augmentation = Sequential(
    [
        layers.RandomRotation(factor=0.15),
        layers.RandomTranslation(height_factor=0.1, width_factor=0.1),
        layers.RandomFlip(),
        layers.RandomContrast(factor=0.1),
    ],
    name="img_augmentation",
)
# applying augmentations on the image
for image, label in ds_train.take(1):
    for i in range(9):
        ax = plt.subplot(3, 3, i + 1)
        aug_img = img_augmentation(tf.expand_dims(image, axis=0)) #augmentation sequential layer
        plt.imshow(aug_img[0].numpy().astype("uint8")) #imshow need numpy array with 'unsigned-int8' precision
        plt.title("{}".format(format_label(label)))
        plt.axis("off")

Awesome ways of using callbacks

https://blog.paperspace.com/tensorflow-callbacks/

Pre-trained model with grayscale images

https://stackoverflow.com/questions/51995977/how-can-i-use-a-pre-trained-neural-network-with-grayscale-images

Tensorflow memory issues with colab

tensorflow/tensorflow#31312

cv2.imshow in google colab

from google.colab.patches import cv2_imshow
cv2_imshow("output.png")

https://stackoverflow.com/questions/55288657/image-is-not-displaying-in-google-colab-while-using-imshow

Disable eagar execution in tensorflow

tf.compat.v1.disable_eager_execution()

Color channel swapping for images

https://www.scivision.dev/numpy-image-bgr-to-rgb/

b,g,r = cv2.split("img.jpg")
data_rgb= cv2.merge([r,g,b])

Multi-Label Classification

https://github.com/novoforce/Exploring-Tensorflow/blob/main/multi_label_classification.ipynb

Download files to colab

!wget -O <output file with extension> --no-check-certificate "<download link>"

! curl <download link without quotes> > <output_file>

# If public sharing link is available
!gdown --id 1cewMfusmPjYWbrnuJRuKhPMwRe_b9PaT

If shared link is only for you:

https://stackoverflow.com/questions/62759748/downloading-data-from-a-shared-google-drive-link-in-google-colab

Extra links:

https://stackoverflow.com/questions/48735600/file-download-from-google-drive-to-colaboratory

Download files from colab to local system

!zip -r /content/file.zip /content/Folder_To_Zip

from google.colab import files
files.download("/content/file.zip")

Import python files from different location

import sys 
import os
sys.path.append(os.path.abspath("/home/el/foo4/stuff")) # attach the path of the directory to be included
from riaa import *
watchout()

https://stackoverflow.com/questions/2349991/how-to-import-other-python-files

Unrar, Unzip, Untar Rar, Zip, Tar in GDrive

https://colab.research.google.com/github/sudo-ken/compress-decompress-in-Google-Drive/blob/master/Unrar_Unzip_Rar_Zip_in_GDrive.ipynb#scrollTo=o1PFiGyMAMvg

References for the badges

Tutorial:

https://img.shields.io/static/v1?label=put-custom-label&message=Tutorial&color=yellow&style=flat-square

Snippet:

https://img.shields.io/static/v1?label=put-custom-label&message=Snippet&color=blue&style=flat-square

Issue:

https://img.shields.io/static/v1?label=put-custom-label&message=Issue&color=red&style=flat-square