FAI_Fall24/CNNModel.py at main · anshuman-raina/FAI_Fall24 · 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
from tensorflow.keras import layers, Input
import tensorflow as tf

def build_cnn_model_with_residuals(input_shape=(640, 640, 1)):

    input_tensor = Input(shape=input_shape)


    x = layers.Conv2D(32, (7, 7), activation='relu', padding='same')(input_tensor)
    x = layers.BatchNormalization()(x)
    x = layers.MaxPooling2D(pool_size=(2, 2))(x)


    shortcut = x
    x = layers.Conv2D(64, (5, 5), activation='relu', padding='same')(x)
    x = layers.BatchNormalization()(x)
    x = layers.MaxPooling2D(pool_size=(2, 2))(x)
    shortcut = layers.Conv2D(64, (1, 1), strides=(2, 2), padding='same')(shortcut)
    x = layers.Add()([x, shortcut])


    shortcut = x
    x = layers.Conv2D(128, (3, 3), activation='relu', padding='same')(x)
    x = layers.BatchNormalization()(x)
    x = layers.MaxPooling2D(pool_size=(2, 2))(x)
    shortcut = layers.Conv2D(128, (1, 1), strides=(2, 2), padding='same')(shortcut)
    x = layers.Add()([x, shortcut])


    shortcut = x
    x = layers.Conv2D(256, (3, 3), activation='relu', padding='same')(x)
    x = layers.BatchNormalization()(x)
    x = layers.MaxPooling2D(pool_size=(2, 2))(x)
    shortcut = layers.Conv2D(256, (1, 1), strides=(2, 2), padding='same')(shortcut)
    x = layers.Add()([x, shortcut])


    shortcut = x
    x = layers.Conv2D(512, (3, 3), activation='relu', padding='same')(x)
    x = layers.BatchNormalization()(x)
    x = layers.MaxPooling2D(pool_size=(2, 2))(x)
    shortcut = layers.Conv2D(512, (1, 1), strides=(2, 2), padding='same')(shortcut)
    x = layers.Add()([x, shortcut])


    x = layers.GlobalAveragePooling2D()(x)

    # Fully Connected Layers
    x = layers.Dense(1024, activation='relu')(x)
    x = layers.Dropout(0.5)(x)
    x = layers.Dense(512, activation='relu')(x)
    x = layers.Dropout(0.3)(x)
    x = layers.Dense(256, activation='relu')(x)
    x = layers.Dropout(0.3)(x)

    # Classification Output (Binary: Empty or Not Empty)
    classification_output = layers.Dense(1, activation='sigmoid', name='classification_output')(x)

    # Optional: Bounding Box Regression Output
    bbox_output = layers.Dense(4, name='bbox_output')(x)

    # Create Model
    model = tf.keras.Model(inputs=input_tensor, outputs=[bbox_output, classification_output])

    return model