Starter Project

!ls /datasets/drive

!gdown https://drive.google.com/uc?id=1o9frkoTA0srFBXOD7zPEzh6pbS9KVkdG&export=download !unzip img

#import libraries and packages from numpy.random import seed seed(8) #1 import tensorflow tensorflow.random.set_seed(7) import numpy as np # linear algebra import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv) import os from tensorflow.keras import backend as K from tensorflow.keras.models import Model ,load_model from tensorflow.keras.layers import Flatten, Dense, Dropout from tensorflow.keras.applications.inception_resnet_v2 import InceptionResNetV2, preprocess_input from keras.applications.vgg16 import preprocess_input from keras.applications.vgg16 import decode_predictions from keras.applications.vgg16 import VGG16 from tensorflow.keras.optimizers import Adam, RMSprop from tensorflow.keras.preprocessing.image import ImageDataGenerator from tensorflow.keras.callbacks import ModelCheckpoint import numpy as np import tensorflow as tf from tensorflow.python.keras import models from tensorflow.python.keras import layers from tensorflow.keras import optimizers

from os import listdir #data_list = listdir('/content/covid-19/four_classes/train') data_list = listdir('/datasets/drive/img/train') print(data_list) print(len(data_list))

DATASET_PATH = '/datasets/drive/img/train' test_dir = '/datasets/drive/img/test' IMAGE_SIZE = (150, 150) NUM_CLASSES = len(data_list) BATCH_SIZE = 10 # try reducing batch size or freeze more layers if your GPU runs out of memory NUM_EPOCHS = 80 LEARNING_RATE =0.0001

#Train datagen here is a preprocessor train_datagen = ImageDataGenerator(rescale=1./255, rotation_range=50, featurewise_center = True, featurewise_std_normalization = True, width_shift_range=0.2, height_shift_range=0.2, shear_range=0.25, zoom_range=0.1, zca_whitening = True, channel_shift_range = 20, horizontal_flip = True , vertical_flip = True , validation_split = 0.2, fill_mode='constant')

# For multiclass use categorical n for binary use binary train_batches = train_datagen.flow_from_directory(DATASET_PATH, target_size=IMAGE_SIZE, shuffle=True, batch_size=BATCH_SIZE, subset = "training", seed=42, class_mode="categorical" #For multiclass use categorical n for binary use binary ) valid_batches = train_datagen.flow_from_directory(DATASET_PATH, target_size=IMAGE_SIZE, shuffle=True, batch_size=BATCH_SIZE, subset = "validation", seed=42, class_mode="categorical" #For multiclass use categorical n for binary use binary )

#Simple CNN model based on Xception. Set dense layer neuron count same as the no. of output classes #If you wnna use a saved model then skip this step from tensorflow.keras.applications import Xception conv_base = Xception(weights='imagenet', include_top=False, input_shape=(150, 150, 3)) conv_base.trainable = True #FALSE? model = models.Sequential() model.add(conv_base) model.add(layers.Flatten()) model.add(layers.Dropout(0.5)) model.add(layers.Dense(256, activation='relu')) model.add(layers.Dense(5, activation='softmax')) model.compile(loss='categorical_crossentropy', #for multiclass use categorical_crossentropy optimizer=optimizers.Adam(lr=LEARNING_RATE), metrics=['acc'])

print(model.summary())

model=load_model('/datasets/drive/img/4-class-Covid19-Mod-Xception.h5') print(model.summary())

# Create evaluate data generator from test set #Dont forget shuffle false test_datagen = ImageDataGenerator(rescale=1. / 255) test_dir = '/datasets/drive/img/test' eval_generator = test_datagen.flow_from_directory(test_dir,target_size=IMAGE_SIZE,batch_size=1, shuffle=False, seed=42, class_mode="categorical") eval_generator.reset()

eval_generator.reset()

#Evalute the trained model on evaluate generator #eval_generator.reset() x = model.evaluate_generator(eval_generator, steps = np.ceil(len(eval_generator)), use_multiprocessing = False, verbose = 1, workers=1, ) print('Test loss:' , x[0]) print('Test accuracy:',x[1]) #Poor test accuracy due to the small dataset size

#Create DataGen on single folder/class and predict IMAGE_SIZE = (150, 150) test_datagen = ImageDataGenerator(rescale=1. / 255) test_dir = '/datasets/drive/img/test' pred_generator = test_datagen.flow_from_directory( test_dir,target_size=IMAGE_SIZE, batch_size=1, shuffle=False, seed=42, class_mode="categorical") pred_generator.reset() count=[0,0,0,0,0] files=pred_generator.filenames for i in range(len(files)): x,y = pred_generator.next() img = x predict=model.predict(img) p=np.argmax(predict, axis=-1) print(str(p[0])+" "+files[pred_generator.batch_index-1]) #print(predict) #p=model.predict_classes(img) count[p[0]]+=1 #print(str(p[0])+" "+files[i]) print(count)

#Predict Results using predict generator and evaluate the accuracy and Confusion matrix from sklearn.metrics import confusion_matrix from sklearn.metrics import plot_confusion_matrix from sklearn.metrics import classification_report import matplotlib.font_manager as fm import itertools filenames = eval_generator.filenames nb_samples = len(filenames) print(filenames)

#Predict Results using predict generator and evaluate the accuracy and Confusion matrix from sklearn.metrics import classification_report, confusion_matrix import matplotlib.pyplot as plt #import matplotlib.font_manager as fm import itertools filenames = eval_generator.filenames nb_samples = len(filenames) eval_generator.reset() predict = model.predict_generator(eval_generator,steps = np.ceil(len(eval_generator))) pp=predict predict=np.argmax(predict, axis=-1) #y_pred classes= eval_generator.classes[eval_generator.index_array] #print(classes,[]) acc=sum(predict==classes)/len(predict) names=["aleatorio","covid","normal","pneumonia_bac","pneumonia_vir"] #print(confusion_matrix(classes,predict)) font = { 'family': 'Times New Roman', 'size': 12 } plt.rc('font', **font) def plot_confusion_matrix(cm, classes, normalize=False, title='Confusion matrix', cmap=plt.cm.Oranges): """ This function prints and plots the confusion matrix. Normalization can be applied by setting `normalize=True`. """ if normalize: cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis] print("Normalized confusion matrix") else: print('Confusion matrix, without normalization') print(cm) plt.imshow(cm, interpolation='nearest', cmap=cmap) plt.title(title) plt.colorbar() tick_marks = np.arange(len(classes)) plt.xticks(tick_marks, classes, rotation=45) plt.yticks(tick_marks, classes) fmt = '.2f' if normalize else 'd' thresh = cm.max() / 2. for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])): plt.text(j, i, format(cm[i, j], fmt), horizontalalignment="center", color="white" if cm[i, j] > thresh else "black") plt.tight_layout() plt.ylabel('True label') plt.xlabel('Predicted label')

# Compute confusion matrix cnf_matrix = confusion_matrix(classes, predict) np.set_printoptions(precision=2) print (classification_report(classes, predict))

# Plot non-normalized confusion matrix plt.figure() plot_confusion_matrix(cnf_matrix, classes=['aleatorio(0)','covid(1)','normal(2)','bacteriana(3)','viral(4)'],normalize= False, title='Confusion matrix')

from getpass import getpass uplink_key = getpass('Enter your Uplink key: ') #VXL6JBIGT3DFSRD5JTSQFQTN-266GD4RAMZWARNXS

!pip install anvil-uplink

import anvil.server

anvil.server.connect("VXL6JBIGT3DFSRD5JTSQFQTN-266GD4RAMZWARNXS")

!apt update !apt install ffmpeg libsm6 libxext6 -y

!pip install google.colab

#Test Single image import numpy as np import cv2 as cv from keras.preprocessing.image import ImageDataGenerator, load_img, img_to_array from keras.utils import to_categorical from sklearn.model_selection import train_test_split import matplotlib.pyplot as plt import random import PIL import os #from google.colab.patches import cv2_imshow #from scikit import transform from skimage import transform import matplotlib.pylab as plt import numpy as np import pandas as pd from keras.preprocessing.image import ImageDataGenerator, load_img, img_to_array import anvil.media

@anvil.server.callable def predict_neumonia(file): with anvil.media.TempFile(file) as f: #img_r = load_img(file) img_r = cv.imread(f) img1 = np.array(img_r).astype('float32')/255 img2 = transform.resize(img1, (150, 150, 3)) #plt.imshow(img2) img = np.expand_dims(img2, axis=0) r=model.predict(img) labels={'ALEATORIO':0, 'COVID':1, 'NORMAL':2, 'BACTERIANA':3, 'VIRAL':4 } names= dict((v,k) for k,v in labels.items()) index = np.argmax(r) name=names.get(index,"Unknown") p=round(r.max()*100, 3) #to find maximum score scores=r mayor=0; index=0; for i in range(5): if (round(scores[0][i]*100,1)>mayor): mayor=round(scores[0][i]*100,1); index=i; return names[1] +" "+ str(round(scores[0][1]*100,1))+"%"+"\n"+names[2] +" "+ str(round(scores[0][2]*100,1))+"%"+"\n"+names[3] +" "+ str(round(scores[0][3]*100,1))+"%"+"\n"+names[4] +" "+ str(round(scores[0][4]*100,1))+"%", index

anvil.server.wait_forever()