I trained ResNet-50 model to classify images from 6 classes (my own dataset) and saved it. But the model did not learn properly and predictions are incorrect. What would be the reason for this poor learning?
Here is my code, and the output plots using Keras and TensorFlow backend. How can I solve this?
from keras.applications.resnet50 import ResNet50, preprocess_input
from keras.layers import Dense, Dropout
from keras.models import Model
from keras.optimizers import Adam, SGD
from keras.preprocessing.image import ImageDataGenerator, image
from keras.callbacks import EarlyStopping, ModelCheckpoint
from sklearn.metrics import confusion_matrix, classification_report, accuracy_score
from keras import backend as K
import numpy as np
import matplotlib.pyplot as plt
from PIL import ImageFile
ImageFile.LOAD_TRUNCATED_IMAGE = True
# Define some constant needed throughout the script
N_CLASSES = 6
EPOCHS = 20
PATIENCE = 5
TRAIN_PATH= '/Train/'
VALID_PATH = '/Test/'
MODEL_CHECK_WEIGHT_NAME = 'resnet_monki_v1_chk.h5'
# Define model to be used we freeze the pre trained resnet model weight, and add few layer on top of it to utilize our custom dataset
K.set_learning_phase(0)
model = ResNet50(input_shape=(224,224,3),include_top=False, weights='imagenet', pooling='avg')
K.set_learning_phase(1)
x = model.output
x = Dense(512, activation='relu')(x)
x = Dropout(0.5)(x)
x = Dense(512, activation='relu')(x)
x = Dropout(0.5)(x)
output = Dense(N_CLASSES, activation='softmax', name='custom_output')(x)
custom_resnet = Model(inputs=model.input, outputs = output)
for layer in model.layers:
layer.trainable = False
custom_resnet.compile(Adam(lr=0.001), loss='categorical_crossentropy', metrics=['accuracy'])
custom_resnet.summary()
# 4. Load dataset to be used
datagen = ImageDataGenerator(preprocessing_function=preprocess_input)
traingen = datagen.flow_from_directory(TRAIN_PATH, target_size=(224,224), batch_size=32, class_mode='categorical')
validgen = datagen.flow_from_directory(VALID_PATH, target_size=(224,224), batch_size=32, class_mode='categorical', shuffle=False)
# 5. Train Model we use ModelCheckpoint to save the best model based on validation accuracy
es_callback = EarlyStopping(monitor='val_acc', patience=PATIENCE, mode='max')
mc_callback = ModelCheckpoint(filepath=MODEL_CHECK_WEIGHT_NAME, monitor='val_acc', save_best_only=True, mode='max')
train_history = custom_resnet.fit_generator(traingen, steps_per_epoch=len(traingen), epochs= EPOCHS, validation_data=traingen, validation_steps=len(validgen), verbose=2, callbacks=[es_callback, mc_callback])
custom_resnet.save('custom_resnet.h5')
Here are the plots, I had to put the links, the site does not let me put a pic
enter image description here
Related
I have trained my model(binary classification) of CNN using Keras and now I want to use an SVM classifier instead of using a fully connected layer for classification.
I used a VGG16 pre-trained network for feature extraction and I also used Data augmentation.
What is the possible way to add SVM as the last layer for classification?
#Parametres
import keras
from keras.applications import VGG16
import sys
from PIL import Image
#Using VGG16 Pre-trained Model
conv_base = VGG16(weights = 'imagenet',
include_top = False,
input_shape=(224, 224, 3))
conv_base.summary()
import numpy as np
import os
from keras.preprocessing.image import ImageDataGenerator
base_dir = 'C:Covid Detection/Code/Dataset-created')
train_dir = os.path.join(base_dir, 'train')
validation_dir = os.path.join(base_dir, 'validation')
test_dir = os.path.join(base_dir, 'test')
from keras import models
from keras import layers
model = models.Sequential()
model.add(conv_base)
model.add(layers.Flatten())
model.add(layers.Dense(256, activation='relu'))
model.add(layers.Dense(1, activation='sigmoid'))
conv_base.trainable = False
from keras import optimizers
train_datagen = ImageDataGenerator(
rescale=1./255,
rotation_range=40,
width_shift_range=0.2,
height_shift_range=0.2,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True,
fill_mode='nearest')
test_datagen = ImageDataGenerator(rescale=1./255)
train_generator = train_datagen.flow_from_directory(
train_dir,
target_size=(224, 224),
batch_size=20,
class_mode='binary')
validation_generator = test_datagen.flow_from_directory(
validation_dir,
target_size=(224, 224),
batch_size=20,
class_mode='binary')
# Compile the model
model.compile(loss='binary_crossentropy',
optimizer=optimizers.RMSprop(lr=2e-5),
metrics=['acc'])
# Train the model
history = model.fit_generator(
train_generator,
steps_per_epoch=50,
epochs=30,
validation_data=validation_generator,
validation_steps=50)
# Save the model
model.save('vgg16_aug.h5')
from keras.applications.vgg16 import VGG16
from keras.preprocessing.image import load_img
from keras.preprocessing.image import img_to_array
from keras.applications.vgg16 import preprocess_input
base_model = VGG16(weights='imagenet')
model = Model(inputs=base_model.input,
outputs=base_model.get_layer('flatten').output)
model.summary()
Create a function to extract features using VGG16
def get_features(img_path):
img = load_img(img_path, target_size=(224, 224))
x = img_to_array(img)
x = np.expand_dims(x, axis=0)
x = preprocess_input(x)
flatten = model.predict(x)
return list(flatten[0])
features, labels = [], []
# Loop into the directory of images and extract features and labels
for image_path in folder:
features.append(extract_features(image_path))
labels.append("#some label")
Till here we extracted the features and labels, Now we train these features using SVM as follows:-
from sklearn.model_selection import train_test_split
from sklearn.svm import LinearSVC
from sklearn.metrics import accuracy_score
X_train, X_test, y_train, y_test = train_test_split( features,
labels,
test_size=0.30)
clf = LinearSVC(random_state=0, tol=1e-5)
clf.fit(X_train, y_train)
predicted = clf.predict(X_test)
# get the accuracy
print(accuracy_score(y_test, predicted))
When I try to use the TimeSeriesGenerator function, my Keras LSTM NN starts training for a few moments but then gives a ValueError message. What's wrong? I wonder how it can start training and then get an error.
My similar implementation without this function runs smoothly but then the quality of the predictions are awful (and I'm not sure that this function, once successfully implemented, would make a difference).
See the code below:
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from tensorflow.keras.models import Model
from tensorflow.keras.optimizers import Nadam
from tensorflow.keras.layers import Input, LSTM, Dense
from tensorflow.keras.callbacks import EarlyStopping, ReduceLROnPlateau, TerminateOnNaN
from tensorflow.keras.preprocessing.sequence import TimeseriesGenerator
data = pd.read_excel('example.xlsx',usecols=['wave','wind','current','X','Y','RZ'])
data = data.apply(lambda x: (x - np.mean(x)) / np.std(x))
n_cutoff = 200
X = np.array(data.loc[n_cutoff:,['wave','wind']])
Y = np.array(data.loc[n_cutoff:,['RZ']])
X = X.reshape(len(X),2)
X = np.append(X, [[0]*np.size(X, axis=1)], axis=0)
Y = Y.reshape(len(Y),1)
Y = np.insert(Y, 0, 0)
n_lag = 3
n_batch = 15
n = int(0.75*len(X))
generator = TimeseriesGenerator(X, Y, length=n_lag, batch_size=n_batch)
inputs = Input(shape=(n_lag,2))
hidden1 = LSTM(units=100,
activation='softmax',
recurrent_activation='linear',
dropout=0.5,
recurrent_dropout=0.5,
return_sequences=True)(inputs)
hidden2 = LSTM(units=30,
activation='softmax',
recurrent_activation='linear',
dropout=0.5,
recurrent_dropout=0.5,
return_sequences=False)(hidden1)
outputs = Dense(units=1,
activation='linear')(hidden2)
model = Model(inputs=inputs, outputs=outputs)
optimizer = Nadam(learning_rate=1e-2, beta_1=0.95, beta_2=0.9, epsilon=1e-7)
model.compile(loss='mean_squared_error', optimizer=optimizer)
history = model.fit(generator,
verbose=1,
steps_per_epoch=int(n/n_batch),
epochs=1,
shuffle=False,
callbacks=[EarlyStopping(monitor='loss', min_delta=0, patience=20, verbose=1, mode='auto'),
ReduceLROnPlateau(monitor='loss', factor=0.5, patience=10, verbose=1, mode='auto', cooldown=1),
TerminateOnNaN()])
Y_hat = model.predict(X[n:])
I am using the following code to apply sequential LSTM to time-series data with one value. It works fine with a Keras version. I am wondering how could I do the same using PyTorch?
import tensorflow
from tensorflow.keras import optimizers
from tensorflow.keras import losses
from tensorflow.keras.models import Sequential, Model
from tensorflow.keras.layers import Dense, Input, Dropout, Embedding, LSTM
from tensorflow.keras.optimizers import RMSprop, Adam, Nadam
from tensorflow.keras.preprocessing import sequence
from tensorflow.keras.callbacks import TensorBoard
# training_dataset.shape = (303, 24, 1)
time_steps = 24
metric = 'mean_absolute_error'
model = Sequential()
model.add(LSTM(units=32, activation='tanh', input_shape=(time_steps, 1), return_sequences=True))
model.add(Dense(1, activation='sigmoid'))
model.compile(optimizer='adam', loss='mean_absolute_error', metrics=[metric])
print(model.summary())
batch_size=32
epochs=20
model.fit(x=training_dataset, y=training_dataset,
batch_size=batch_size, epochs=epochs,
verbose=1, validation_data=(training_dataset, training_dataset),
callbacks=[TensorBoard(log_dir='../logs/{0}'.format(tensorlog))])
testing_pred = model.predict(x=testing_dataset)
You can check the pytorch documentation for that: https://pytorch.org/docs/master/generated/torch.nn.LSTM.html
the simplest code is the following:
import torch, torch.nn as nn, torch.optim.Adam as Adam
model = nn.Sequential(nn.LSTM(input_size=1, hidden_size=32, output_size=1), nn.Sigmoid)
opt = Adam(model.parameters())
loss_func = nn.MSELoss()
for (x, y) in dataloader:
opt.zero_grad()
pred = model(x)
loss = loss_func(y, pred)
loss.backward()
opt.step()
I loaded Fahion_Mnist dataset through "fashion_mnist.load_data()" and I tried to train a ResNet50 neural network. But I don't know how reshape dataset image from (28,28,1) to (224,224,3), as needed as input in ResNet.
I am using Python 3, Keras 2.2.4
This is my code:
from __future__ import absolute_import, division, print_function
import tensorflow as tf
from tensorflow import keras# Helper libraries
import numpy as np
import matplotlib.pyplot as plt
import time
from tensorflow.python.keras import backend as K
from tensorflow.python.keras.models import Model
from tensorflow.python.keras.layers import Flatten, Dense, Dropout
from tensorflow.python.keras.applications.resnet50 import ResNet50, preprocess_input
from tensorflow.python.keras.optimizers import Adam
from tensorflow.python.keras.preprocessing.image import ImageDataGenerator
from tensorflow.python.keras.preprocessing import image
from PIL import Image
fashion_mnist = keras.datasets.fashion_mnist
(train_images, train_labels), (test_images, test_labels) = fashion_mnist.load_data()
class_names = ['T-shirt/top', 'Trouser', 'Pullover', 'Dress', 'Coat','Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle boot']
IMAGE_SIZE = (224,224)
NUM_CLASSES = 10
BATCH_SIZE = 8 # try reducing batch size or freeze more layers if your GPU runs out of memory
FREEZE_LAYERS = 2 # freeze the first this many layers for training
NUM_EPOCHS = 20
WEIGHTS_FINAL = 'model_fashion_resnet.h5'
train_images = preprocess_input(train_images)
train_images = np.expand_dims(train_images, axis=0)
train_labels = preprocess_input(train_labels)
train_labels = np.expand_dims(train_labels, axis=0)
test_images = preprocess_input(test_images)
test_images = np.expand_dims(test_images, axis=0)
net = ResNet50(include_top=False, weights='imagenet', input_tensor=None,
input_shape=(IMAGE_SIZE[0],IMAGE_SIZE[1],3))
x = net.output
x = Flatten()(x)
x = Dropout(0.5)(x)
output_layer = Dense(NUM_CLASSES, activation='softmax', name='softmax')(x)
model = Model(inputs=net.input, outputs=output_layer)
for layer in model.layers[:FREEZE_LAYERS]:
layer.trainable = False
for layer in model.layers[FREEZE_LAYERS:]:
layer.trainable = True
model.compile(optimizer=Adam(lr=1e-5), loss='categorical_crossentropy', metrics=['accuracy'])
print(model.summary())
inizio=time.time()
datagen = ImageDataGenerator(
featurewise_center=True,
featurewise_std_normalization=True,
rotation_range=20,
width_shift_range=0.2,
height_shift_range=0.2,
horizontal_flip=True)
model.fit_generator(datagen.flow(train_images, train_labels, batch_size=BATCH_SIZE),
steps_per_epoch=len(train_images) / BATCH_SIZE, epochs=NUM_EPOCHS)
And this is what I receive after run:
ValueError: Error when checking input: expected input_1 to have shape (224, 224, 3) but got array with shape (60000, 28, 28)
How to change MNIST images so that they can input in the ResNet50 neural network?
this keras is a simple keras model, I use relu for simplicity and power of learning
also mean_square as a loss, but the model does not fit
I spect the output to be a number [0-9], im not using to_categorical, thas why there is only a output neuron
from keras.datasets import mnist
(data, labels),(x_test,y_test) = mnist.load_data()
from keras.layers import Input, Dense, Flatten
from keras.models import Model
from keras.losses import mean_squared_error
# This returns a tensor
inputs = Input(shape=(28, 28,))
x = Flatten()(inputs)
x = Dense(64, activation='relu')(x)
x = Dense(64, activation='relu')(x)
predictions = Dense(1, activation='relu')(x)
model = Model(inputs=inputs, outputs=predictions)
model.compile(optimizer='rmsprop',
loss=mean_squared_error,
metrics=['accuracy'])
model.fit(data, labels, batch_size=10, epochs=10 ,
validation_data=(x_test, y_test)) # starts training