-
Notifications
You must be signed in to change notification settings - Fork 0
/
lrTrainCarsBrandInception_v3CALLBACK_1_49.py
145 lines (107 loc) · 4.96 KB
/
lrTrainCarsBrandInception_v3CALLBACK_1_49.py
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
143
144
145
# -*- coding: utf-8 -*-
"""
Adapted from
https://medium.com/analytics-vidhya/top-4-pre-trained-models-for-image-classification-with-python-code-a3cb5846248b
by Alfonso Blanco García , Jul 2023
"""
######################################################################
# PARAMETERS
######################################################################
#batch_size = 128
#epochs = 30
######################################################################
import tensorflow as tf
import keras
from keras.preprocessing.image import ImageDataGenerator
from keras.layers import Input, Flatten, Dense, Dropout
from keras.models import Model
from keras import optimizers
from tensorflow.keras.callbacks import EarlyStopping, CSVLogger, ReduceLROnPlateau
import numpy as np
import time
import functools
"""
Descarga del fichero de standford
https://www.kaggle.com/datasets/jessicali9530/stanford-cars-dataset/code?resource=download
https://www.analyticslane.com/2020/02/14/leer-y-guardar-archivos-de-matlab-en-python/#:~:text=La%20lectura%20de%20los%20archivos,archivo%20a%20la%20funci%C3%B3n%20loadmat%20.
DESCARGA ANNOTATIONS Y CLAS NAMES en CSV
https://github.com/BotechEngineering/StanfordCarsDatasetCSV/blob/main/cardatasettest.csv
"""
## image path
train_data_dir = 'KaggleCarsByBrands_1_49\\train'
validation_data_dir = 'KaggleCarsByBrands_1_49\\valid'
## other
img_width, img_height = 224, 224
batch_size = 20
nb_classes = 49
nb_epoch =50
# Callback Settings:
early_stopping_patience = 10
reduce_lr_on_plateau_factor = 0.2
reduce_lr_on_plateau_patience = 3
# https://towardsdatascience.com/transfer-learning-in-action-from-imagenet-to-tiny-imagenet-b96fe3aa5973
#lr_callback — Reduces the learning rate of the optimizer by a factor of 0.1 if the val_loss does not go down within 5 epochs.
lr_callback = ReduceLROnPlateau(
monitor='val_loss', factor=0.1, patience=2, verbose=1, mode='auto'
)
# https://github.com/afaq-ahmad/Car-Models-and-Make-Classification-Standford_Car_dataset-mobilenetv2-imagenet-93-percent-accuracy/blob/master/Car_classification.ipynb
def get_callbacks_list(_early_stopping_patience, _reduce_lr_on_plateau_factor, _reduce_lr_on_plateau_patience, _lr_callback):
"""Get callbacks for a model"""
return [
keras.callbacks.ModelCheckpoint(
verbose=1,
filepath='lrbest_brand_1_49.h5',
mode='max',
monitor='val_acc',
#monitor='val_loss',
#monitor='top6_acc',
save_best_only=True
),
]
# start measurement
start = time.time()
# IGNORE data-augmentation parameters to ImageDataGenerator
#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)
train_datagen = ImageDataGenerator(rescale = 1./255.)
validation_datagen = ImageDataGenerator( rescale = 1.0/255. )
train_generator = train_datagen.flow_from_directory(
train_data_dir,
target_size=(img_width, img_height),
color_mode='rgb',
class_mode='categorical',
batch_size=batch_size,
shuffle=True
)
validation_generator = validation_datagen.flow_from_directory(
validation_data_dir,
target_size=(img_width, img_height),
color_mode='rgb',
class_mode='categorical',
batch_size=batch_size,
shuffle=True
)
from tensorflow.keras.applications.inception_v3 import InceptionV3
base_model = InceptionV3(input_shape = (224, 224, 3), include_top = False, weights = 'imagenet')
#base_model = InceptionV3(input_shape = (224, 224, 3), include_top = False)
#IGNORE only change the last layer.
#for layer in base_model.layers:
# layer.trainable = False
from tensorflow.keras.optimizers import RMSprop
x = Flatten()(base_model.output)
x = Dense(4096, activation='relu')(x)
x = Dropout(0.2)(x)
x = Dense(3072, activation='relu')(x)
x = Dropout(0.2)(x)
x = Dense(2048, activation='relu')(x)
x = Dropout(0.2)(x)
x = Dense(1024, activation='relu')(x)
x = Dropout(0.2)(x)
x = Dense(nb_classes, activation='softmax')(x)
model = tf.keras.models.Model(base_model.input, x)
#model.compile(optimizer = RMSprop(learning_rate=0.0001), loss = 'categorical_crossentropy', metrics = ['acc'])
model.compile(optimizer = RMSprop(learning_rate=0.00005), loss = 'categorical_crossentropy', metrics = ['acc'])
#model.compile(optimizer = optimizers.SGD(lr=0.0001), loss = 'categorical_crossentropy', metrics = ['acc'])
inc_history = model.fit(train_generator, validation_data = validation_generator,
steps_per_epoch = 10, epochs = nb_epoch,
callbacks=get_callbacks_list(early_stopping_patience, reduce_lr_on_plateau_factor, reduce_lr_on_plateau_patience, lr_callback))
model.save("lrModelCarsBrands_Inception_v3_1_49.h5")