A simple lightweight attention mechanism wrapper in 2D convolution. It consists with attentin mdoules of Convolution Block Attention Module (CBAM) and DeepMoji.
CBAM: Convolutional Block Attention Module is a dual attention
mechanism. It learns the informative features by integrating channel-wise attention and spatial attention together. The module is set in sequential order, begin with channel-wise followed by the spatial module. Additionally we add a weighting term to the Global Average Pooling (GAP) at the end of CBAM; such as
simply run
python train.py
Epoch 1/5
1875/1875 [==============================] - 178s 89ms/step - loss: 0.4090 - accuracy: 0.8990
Epoch 2/5
1875/1875 [==============================] - 166s 89ms/step - loss: 0.0971 - accuracy: 0.9744
Epoch 3/5
1875/1875 [==============================] - 166s 89ms/step - loss: 0.0910 - accuracy: 0.9775
Epoch 4/5
1875/1875 [==============================] - 166s 89ms/step - loss: 0.0797 - accuracy: 0.9810
Epoch 5/5
1875/1875 [==============================] - 166s 89ms/step - loss: 0.0611 - accuracy: 0.9851
Model details
class Classifier(tf.keras.Model):
def __init__(self, dim):
super(Classifier, self).__init__()
# Defining All Layers in __init__
# Layer of Block
self.Base = tf.keras.applications.Xception(
input_shape=(image_size, image_size, 3),
weights=None,
include_top=False)
# Keras Built-in
self.GAP1 = tf.keras.layers.GlobalAveragePooling2D()
self.GAP2 = tf.keras.layers.GlobalAveragePooling2D()
self.BAT = tf.keras.layers.BatchNormalization()
self.ADD = tf.keras.layers.Add()
self.AVG = tf.keras.layers.Average()
self.DROP = tf.keras.layers.Dropout(rate=0.5)
# Customs
self.CAN = ChannelAttentionModule()
self.SPN1 = SpatialAttentionModule()
self.SPN2 = SpatialAttentionModule()
self.AWG = AttentionWeightedAverage2D()
# Tail
self.DENS = tf.keras.layers.Dense(512, activation=tf.nn.relu)
self.OUT = tf.keras.layers.Dense(10, activation='softmax', dtype=tf.float32)
def call(self, input_tensor, training=False):
# Base Inputs
x = self.Base(input_tensor)
# Attention Modules 1
# Channel Attention + Spatial Attention
canx = self.CAN(x)*x
spnx = self.SPN1(canx)*canx
# Global Weighted Average Poolin
gapx = self.GAP1(spnx)
wvgx = self.GAP2(self.SPN2(canx))
gapavg = self.AVG([gapx, wvgx])
# Attention Modules 2
# Attention Weighted Average (AWG)
awgavg = self.AWG(x)
# Summation of Attentions
x = self.ADD([gapavg, awgavg])
# Tails
x = self.BAT(x)
x = self.DENS(x)
x = self.DROP(x, training=training)
return self.OUT(x)
For live example, check this notebook.