Skip to content

Instantly share code, notes, and snippets.

@jainxy

jainxy/models_def_samples_pub.py

Created March 3, 2021 04:41
Show Gist options
  • Save jainxy/9f54e4cf94d05e1dc90dad4b4b6d5348 to your computer and use it in GitHub Desktop.
Save jainxy/9f54e4cf94d05e1dc90dad4b4b6d5348 to your computer and use it in GitHub Desktop.
Download ZIP
Model architecture samples
Raw
models_def_samples_pub.py
"""
A collection of models we'll use to attempt to classify videos.
"""
from keras.layers import Dense, Flatten, Dropout, ZeroPadding3D
from keras.layers.recurrent import LSTM
from keras.models import Sequential, load_model
from keras.optimizers import Adam, RMSprop
from keras.layers.wrappers import TimeDistributed
from keras.layers.convolutional import (Conv2D, MaxPooling3D, Conv3D,
MaxPooling2D)
from collections import deque
import sys
class TraialModels():
def __init__(self, nb_classes, model, seq_length,
saved_model=None, features_length=2048):
"""
`model` = one of:
lstm
lrcn
mlp
conv_3d
c3d
`nb_classes` = the number of classes to predict
`seq_length` = the length of our video sequences
`saved_model` = the path to a saved Keras model to load
"""
# Set defaults.
self.seq_length = seq_length
self.load_model = load_model
self.saved_model = saved_model
self.nb_classes = nb_classes
self.feature_queue = deque()
# Set the metrics. Only use top k if there's a need.
metrics = ['accuracy']
if self.nb_classes >= 10:
metrics.append('top_k_categorical_accuracy')
# Get the appropriate model.
if self.saved_model is not None:
print("Loading model %s" % self.saved_model)
self.model = load_model(self.saved_model)
elif model == 'lstm':
print("Loading LSTM model.")
self.input_shape = (seq_length, features_length)
self.model = self.lstm()
elif model == 'lrcn':
print("Loading CNN-LSTM model.")
self.input_shape = (seq_length, 80, 80, 3)
self.model = self.lrcn()
elif model == 'mlp':
print("Loading simple MLP.")
self.input_shape = (seq_length, features_length)
self.model = self.mlp()
elif model == 'conv_3d':
print("Loading Conv3D")
self.input_shape = (seq_length, 80, 80, 3)
self.model = self.conv_3d()
elif model == 'c3d':
print("Loading C3D")
self.input_shape = (seq_length, 80, 80, 3)
self.model = self.c3d()
else:
print("Unknown network.")
sys.exit()
# Now compile the network.
optimizer = Adam(lr=1e-5, decay=1e-6)
self.model.compile(loss='categorical_crossentropy', optimizer=optimizer,
metrics=metrics)
print(self.model.summary())
def lstm(self):
"""Simple LSTM network."""
# Model.
model = Sequential()
model.add(LSTM(2048, return_sequences=False,
input_shape=self.input_shape,
dropout=0.5))
model.add(Dense(512, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(self.nb_classes, activation='softmax'))
return model
def lrcn(self):
def add_default_block(model, kernel_filters, init, reg_lambda):
# conv
model.add(TimeDistributed(Conv2D(kernel_filters, (3, 3), padding='same',
kernel_initializer=init, kernel_regularizer=L2_reg(l=reg_lambda))))
model.add(TimeDistributed(BatchNormalization()))
model.add(TimeDistributed(Activation('relu')))
# conv
model.add(TimeDistributed(Conv2D(kernel_filters, (3, 3), padding='same',
kernel_initializer=init, kernel_regularizer=L2_reg(l=reg_lambda))))
model.add(TimeDistributed(BatchNormalization()))
model.add(TimeDistributed(Activation('relu')))
# max pool
model.add(TimeDistributed(MaxPooling2D((2, 2), strides=(2, 2))))
return model
initialiser = 'glorot_uniform'
reg_lambda = 0.001
model = Sequential()
# first (non-default) block
model.add(TimeDistributed(Conv2D(32, (7, 7), strides=(2, 2), padding='same',
kernel_initializer=initialiser, kernel_regularizer=L2_reg(l=reg_lambda)),
input_shape=self.input_shape))
model.add(TimeDistributed(BatchNormalization()))
model.add(TimeDistributed(Activation('relu')))
model.add(TimeDistributed(Conv2D(32, (3,3), kernel_initializer=initialiser, kernel_regularizer=L2_reg(l=reg_lambda))))
model.add(TimeDistributed(BatchNormalization()))
model.add(TimeDistributed(Activation('relu')))
model.add(TimeDistributed(MaxPooling2D((2, 2), strides=(2, 2))))
# 2nd-5th (default) blocks
model = add_default_block(model, 64, init=initialiser, reg_lambda=reg_lambda)
model = add_default_block(model, 128, init=initialiser, reg_lambda=reg_lambda)
model = add_default_block(model, 256, init=initialiser, reg_lambda=reg_lambda)
model = add_default_block(model, 512, init=initialiser, reg_lambda=reg_lambda)
# LSTM output head
model.add(TimeDistributed(Flatten()))
model.add(LSTM(256, return_sequences=False, dropout=0.5))
model.add(LSTM(256, return_sequences=False, dropout=0.5))
model.add(LSTM(256, return_sequences=False, dropout=0.5))
model.add(Dense(self.nb_classes, activation='softmax'))
return model
def mlp(self):
"""A simple MLP."""
# Model.
model = Sequential()
model.add(Flatten(input_shape=self.input_shape))
model.add(Dense(512))
model.add(Dropout(0.5))
model.add(Dense(512))
model.add(Dropout(0.5))
model.add(Dense(self.nb_classes, activation='softmax'))
return model
def conv_3d(self):
# Model.
model = Sequential()
model.add(Conv3D(
32, (3,3,3), activation='relu', input_shape=self.input_shape
))
model.add(MaxPooling3D(pool_size=(1, 2, 2), strides=(1, 2, 2)))
model.add(Conv3D(64, (3,3,3), activation='relu'))
model.add(MaxPooling3D(pool_size=(1, 2, 2), strides=(1, 2, 2)))
model.add(Conv3D(128, (3,3,3), activation='relu'))
model.add(Conv3D(128, (3,3,3), activation='relu'))
model.add(MaxPooling3D(pool_size=(1, 2, 2), strides=(1, 2, 2)))
model.add(Conv3D(256, (2,2,2), activation='relu'))
model.add(Conv3D(256, (2,2,2), activation='relu'))
model.add(MaxPooling3D(pool_size=(1, 2, 2), strides=(1, 2, 2)))
model.add(Flatten())
model.add(Dense(1024))
model.add(Dropout(0.5))
model.add(Dense(1024))
model.add(Dropout(0.5))
model.add(Dense(self.nb_classes, activation='softmax'))
return model
def c3d(self):
model = Sequential()
# 1st layer group
model.add(Conv3D(64, 3, 3, 3, activation='relu',
border_mode='same', name='conv1',
subsample=(1, 1, 1),
input_shape=self.input_shape))
model.add(MaxPooling3D(pool_size=(1, 2, 2), strides=(1, 2, 2),
border_mode='valid', name='pool1'))
# 2nd layer group
model.add(Conv3D(128, 3, 3, 3, activation='relu',
border_mode='same', name='conv2',
subsample=(1, 1, 1)))
model.add(MaxPooling3D(pool_size=(2, 2, 2), strides=(2, 2, 2),
border_mode='valid', name='pool2'))
# 3rd layer group
model.add(Conv3D(256, 3, 3, 3, activation='relu',
border_mode='same', name='conv3a',
subsample=(1, 1, 1)))
model.add(Conv3D(256, 3, 3, 3, activation='relu',
border_mode='same', name='conv3b',
subsample=(1, 1, 1)))
model.add(MaxPooling3D(pool_size=(2, 2, 2), strides=(2, 2, 2),
border_mode='valid', name='pool3'))
# 4th layer group
model.add(Conv3D(512, 3, 3, 3, activation='relu',
border_mode='same', name='conv4a',
subsample=(1, 1, 1)))
model.add(Conv3D(512, 3, 3, 3, activation='relu',
border_mode='same', name='conv4b',
subsample=(1, 1, 1)))
model.add(MaxPooling3D(pool_size=(2, 2, 2), strides=(2, 2, 2),
border_mode='valid', name='pool4'))
# 5th layer group
model.add(Conv3D(512, 3, 3, 3, activation='relu',
border_mode='same', name='conv5a',
subsample=(1, 1, 1)))
model.add(Conv3D(512, 3, 3, 3, activation='relu',
border_mode='same', name='conv5b',
subsample=(1, 1, 1)))
model.add(ZeroPadding3D(padding=(0, 1, 1)))
model.add(MaxPooling3D(pool_size=(2, 2, 2), strides=(2, 2, 2),
border_mode='valid', name='pool5'))
model.add(Flatten())
# FC layers group
model.add(Dense(4096, activation='relu', name='fc6'))
model.add(Dropout(0.5))
model.add(Dense(4096, activation='relu', name='fc7'))
model.add(Dropout(0.5))
model.add(Dense(self.nb_classes, activation='softmax'))
return model
##================= 3DCNN
model = Sequential()
# 1st layer group
model.add(Conv3D(64, 3, 3, 3, activation='relu',
border_mode='same', name='conv1',
subsample=(1, 1, 1),
input_shape=self.input_shape))
model.add(MaxPooling3D(pool_size=(1, 2, 2), strides=(1, 2, 2),
border_mode='valid', name='pool1'))
# 2nd layer group
model.add(Conv3D(128, 3, 3, 3, activation='relu',
border_mode='same', name='conv2',
subsample=(1, 1, 1)))
model.add(MaxPooling3D(pool_size=(2, 2, 2), strides=(2, 2, 2),
border_mode='valid', name='pool2'))
# 3rd layer group
model.add(Conv3D(256, 3, 3, 3, activation='relu',
border_mode='same', name='conv3a',
subsample=(1, 1, 1)))
model.add(Conv3D(256, 3, 3, 3, activation='relu',
border_mode='same', name='conv3b',
subsample=(1, 1, 1)))
model.add(MaxPooling3D(pool_size=(2, 2, 2), strides=(2, 2, 2),
border_mode='valid', name='pool3'))
# 4th layer group
model.add(Conv3D(512, 3, 3, 3, activation='relu',
border_mode='same', name='conv4a',
subsample=(1, 1, 1)))
model.add(Conv3D(512, 3, 3, 3, activation='relu',
border_mode='same', name='conv4b',
subsample=(1, 1, 1)))
model.add(MaxPooling3D(pool_size=(2, 2, 2), strides=(2, 2, 2),
border_mode='valid', name='pool4'))
# 5th layer group
model.add(Conv3D(512, 3, 3, 3, activation='relu',
border_mode='same', name='conv5a',
subsample=(1, 1, 1)))
model.add(Conv3D(512, 3, 3, 3, activation='relu',
border_mode='same', name='conv5b',
subsample=(1, 1, 1)))
model.add(ZeroPadding3D(padding=(0, 1, 1)))
model.add(MaxPooling3D(pool_size=(2, 2, 2), strides=(2, 2, 2),
border_mode='valid', name='pool5'))
model.add(GlobalAveragePooling3D())
# FC layers group
model.add(Dense(4096, activation='relu', name='fc6'))
model.add(Dropout(0.5))
model.add(Dense(4096, activation='relu', name='fc7'))
model.add(Dropout(0.5))
model.add(Dense(self.nb_classes, activation='softmax'))
return model
## 704x396 / maintain aspect ratio
## 3DCNN
def get_model(width=128, height=128, depth=64):
"""Build a 3D convolutional neural network model."""
inputs = keras.Input((width, height, depth, 1))
x = layers.Conv3D(filters=64, kernel_size=3, activation="relu")(inputs)
x = layers.MaxPool3D(pool_size=2)(x)
x = layers.BatchNormalization()(x)
x = layers.Conv3D(filters=64, kernel_size=3, activation="relu")(x)
x = layers.MaxPool3D(pool_size=2)(x)
x = layers.BatchNormalization()(x)
x = layers.Conv3D(filters=128, kernel_size=3, activation="relu")(x)
x = layers.MaxPool3D(pool_size=2)(x)
x = layers.BatchNormalization()(x)
x = layers.Conv3D(filters=256, kernel_size=3, activation="relu")(x)
x = layers.MaxPool3D(pool_size=2)(x)
x = layers.BatchNormalization()(x)
x = layers.GlobalAveragePooling3D()(x)
x = layers.Dense(units=512, activation="relu")(x)
x = layers.Dropout(0.3)(x)
outputs = layers.Dense(units=1, activation="sigmoid")(x)
# Define the model.
model = keras.Model(inputs, outputs, name="3dcnn")
return model
# Build model.
model = get_model(width=128, height=128, depth=64)
model.summary()
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment