ecg/train_1d_conv.py at master · embracesource-cv-com/ecg · 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
# -*- coding:utf-8 _*-
"""
@author: danna.li
@date: 2019/3/14
@file: train_ecg.py
@description:
"""
from common.conf import current_config as conf
from deep.base_nets import conv_1d
from keras import Input
from deep.train_common import load_data, compile_model, train_model
import os
import numpy as np
from common import utils

os.environ["CUDA_VISIBLE_DEVICES"] = conf.gpu_index


def choose_model(input_x):
    model_conf = conf.model_1d
    out = None
    if model_conf == 'simple_net':
        out = conv_1d.simple_net(input_x)
    elif model_conf == 'ecg_resnet':
        out = conv_1d.ecg_resnet(input_x)
    elif model_conf == 'mini_resnet':
        out = conv_1d.mini_resnet(input_x)
    return out


def train_lead_as_channel():
    x, y = load_data()
    input_x = Input([conf.seq_len, conf.num_lead])
    out = choose_model(input_x)
    model_compiled = compile_model(input_x, out)
    train_model(x, y, model_compiled)


def lead_to_sample():
    x, y = load_data()
    x, y, x_test, y_test = utils.split_data(x, y, conf.train_ratio)
    x = x.reshape(-1, 5000, 1)
    y = np.repeat(y, 12)
    print('x.shape,y.shape', x.shape, y.shape)
    print('x_test.shape,y_test.shape', x_test.shape, y_test.shape)
    return x, y, x_test, y_test


def train_lead_as_sample():
    x, y, _, _ = lead_to_sample()
    input_x = Input([conf.seq_len, 1])
    out = choose_model(input_x)
    model_compiled = compile_model(input_x, out)
    train_model(x, y, model_compiled)


if __name__ == '__main__':
    if conf.lead_as_sample:
        train_lead_as_sample()
    else:
        train_lead_as_channel()