File size: 1,567 Bytes
e2b7bc1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
from __future__ import annotations
import os
import tensorflow as tf
from tensorflow.keras.layers import (Activation, Bidirectional, Conv3D, Dense,
                                     Dropout, LSTM, MaxPool3D, Reshape)
from tensorflow.keras.models import Sequential

# Disable all GPUS
tf.config.set_visible_devices([], 'GPU')

def load_model() -> Sequential:
    model = Sequential()
    model.add(Conv3D(128, 3, input_shape=(75, 46, 140, 1), padding='same'))
    model.add(Activation('relu'))
    model.add(MaxPool3D((1, 2, 2)))

    model.add(Conv3D(256, 3, padding='same'))
    model.add(Activation('relu'))
    model.add(MaxPool3D((1, 2, 2)))

    model.add(Conv3D(75, 3, padding='same'))
    model.add(Activation('relu'))
    model.add(MaxPool3D((1, 2, 2)))

    # Reshape instead of TimeDistributed(Flatten) — matches your trained weights
    model.add(Reshape((75, 5 * 17 * 75)))

    model.add(Bidirectional(LSTM(128, kernel_initializer='Orthogonal', return_sequences=True)))
    model.add(Dropout(.5))

    model.add(Bidirectional(LSTM(128, kernel_initializer='Orthogonal', return_sequences=True)))
    model.add(Dropout(.5))

    model.add(Dense(41, kernel_initializer='he_normal', activation='softmax'))

    base_dir = os.path.dirname(os.path.abspath(__file__))
    weights_path = os.path.abspath(
        os.path.join(base_dir, '..', 'models', 'checkpoint.weights.h5')
    )
    if not os.path.exists(weights_path):
        raise FileNotFoundError(f"Model weights not found at: {weights_path}")

    model.load_weights(weights_path)
    return model