S3Net/mask_classifier.py at main · myselfbritty/S3Net · GitHub

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
146
147
148
149
150
151
152
153
154
155
#!/usr/bin/env python3
# -*- coding: utf-8 -*-

import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers as LY
import tensorflow.keras.backend as K
tf.keras.backend.set_image_data_format('channels_first')


class ArcFace(LY.Layer):

    def __init__(self, n_classes=783, s=30.0, m=0.50, regularizer=None, **kwargs):
        super(ArcFace, self).__init__(**kwargs)
        self.n_classes = n_classes
        self.s = s
        self.m = m
        self.regularizer = keras.regularizers.get(regularizer)

    def build(self, input_shape):
        super(ArcFace, self).build(input_shape[0])
        self.W = self.add_weight(name='W',
                                shape=(input_shape[0][-1], self.n_classes),
                                initializer='glorot_uniform',
                                trainable=True,
                                regularizer=self.regularizer)

    def call(self, inputs, mode = 'margin', training = True):
        if training:
            if mode == 'margin':
                x, y = inputs
                # normalize feature
                x = tf.nn.l2_normalize(x, axis=-1)
                # normalize weights
                W = tf.nn.l2_normalize(self.W, axis=0)
                # dot product
                logits = x @ W
                # add margin
                # clip logits to prevent zero division when backward
                theta = tf.acos(K.clip(logits, -1.0 + K.epsilon(), 1.0 - K.epsilon()))
                target_logits = tf.cos(theta + self.m)

                logits = logits * (1 - y) + target_logits * y
                # feature re-scale
                logits *= self.s
                out = tf.nn.softmax(logits)
            else:
                x, _ = inputs

                # x = tf.nn.l2_normalize(x, axis=-1)
                # W = tf.nn.l2_normalize(self.W, axis=0)
                logits = x@self.W
                # logits *= self.s
                out = tf.nn.softmax(logits)
        else:
            x, _ = inputs

            # x = tf.nn.l2_normalize(x, axis=-1)
            # W = tf.nn.l2_normalize(self.W, axis=0)
            logits = x@self.W
            # logits *= self.s
            out = tf.nn.softmax(logits)
        return out

    def compute_output_shape(self, input_shape):
        return (None, self.n_classes)


class MultiplyMask(LY.Layer):
    def __init__(self):
        super(MultiplyMask, self).__init__()
        self.ga = LY.GlobalAveragePooling2D()
    def call(self,feature, mask):
        op_features = []

        # for i in range(25):
        for i in range(25):
            current_mask = mask[:,i:i+1,:,:]
            current_feature = tf.multiply(feature, current_mask)
            current_feature = self.ga(current_feature)
            op_features.append(current_feature)
        op = tf.stack(op_features, axis=1)

        return op


class make_model(tf.keras.Model):

    def __init__(self, WEIGHT_DECAY = 1e-4):
        super(make_model, self).__init__()

        self.resnet_model = keras.models.load_model('./pre-trained-weights/Stage_3/resnet_2.h5')
        self.conv1 = LY.Conv2D(256, 1)
        self.conv2 = LY.Conv2D(256, 1)
        self.conv3 = LY.Conv2D(256, 1)
        self.conv4 = LY.Conv2D(256, 1)

        self.up1 = LY.UpSampling2D(size=(8, 8), name="pre_cat_2")
        self.up2 = LY.UpSampling2D(size=(4, 4), name="pre_cat_3")
        self.up3 = LY.UpSampling2D(size=(2, 2), name="pre_cat_4")

        self.concat = LY.Concatenate(axis = 1)
        self.masking_layer  = MultiplyMask()


        self.dense1 = LY.Dense(128, activation = 'relu')

    def call(self, inputs, training = False):
        ip = inputs[0]
        mask_ip = inputs[1]
        op = self.resnet_model(ip)
        stage_2 = op[0]
        stage_3 = op[1]
        stage_4 = op[2]
        stage_5 = op[3]

        stage_5 = self.conv1(stage_5)
        stage_4 = self.conv2(stage_4)
        stage_3 = self.conv3(stage_3)
        stage_2 = self.conv4(stage_2)

        fp_2 = self.up1(stage_5)
        fp_3 = self.up2(stage_4)
        fp_4 = self.up3(stage_3)
        fp_5 = stage_2

        fp_o = self.concat([fp_2, fp_3, fp_4, fp_5])
        fp_o = self.masking_layer(fp_o, mask_ip)
        x = self.dense1(fp_o)

        return x

class make_classifier(tf.keras.Model):

    def __init__(self, num_classes = 10):
        super(make_classifier, self).__init__()
        self.num_classes = num_classes
        self.arc_layer = ArcFace(n_classes=self.num_classes, s=30, m=0.4)

    def call(self, inputs, mode='margin', training=True):
        if mode == 'margin':
            feat = inputs[0]
            y = inputs[1]
        else:
            feat = inputs
            y = tf.zeros(shape = (feat.shape[0], 25, self.num_classes))
        op = self.arc_layer([feat,y], mode = mode, training = training)

        return op