homeworkMarking/train.py at master · LangArthur/homeworkMarking · 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
#!/usr/bin/env python3
#
# Created on Wed Mar 17 2021
#
# Arthur Lang
# train.py
#

import numpy
from tensorflow.keras.datasets import mnist

from src.Model import Model

def to_float(data):
    res=[]
    for item in data:
        res.append(float(item))
    res=numpy.array(res)
    return res


def dataPreprocessing():
    # Mnist dataset
    (dataTrain, labelTrain), (dataTest, labelTest) = mnist.load_data()
    # normalize datas
    maxVal = numpy.max(dataTrain)
    dataTrain = (dataTrain / maxVal).reshape(60000, 28, 28, 1)
    dataTest = (dataTest / maxVal).reshape(10000, 28, 28, 1)
    return dataTrain, labelTrain, dataTest, labelTest

## train
# do a training session, saving the weights
def train():
    dataTrain, labelTrain, dataTest, labelTest = dataPreprocessing()

    nt = Model()
    nt.setOutputWeightPath("ressources/weights/Mnist/")
    nt.train(dataTrain, labelTrain, dataTest, labelTest)
    return 0

if __name__ == "__main__":
    train()