SAComputerVisionMachineLearning/BehavioralCloning/data.py at master · appinho/SAComputerVisionMachineLearning · 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
# Import packages
import csv
import cv2
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.image as mpimg

# define path
directory = '../Recorded_Data/'
data_paths = ['Loop_Data/','Loop_Data_Reversed/',
              'Recover_Data/','Recover_Data_Reversed/',
              'Bridge_Recover_Data/','Curve_Recover_Data/',
              'Recover_Data_2/', 'Loop_Data_2/']
data_usage = [True,True,True,True,True,True,True,True]
data_augmentation = True
use_multiple_cameras = True
correction_factor = 0.2

def load_data():
    images,measurements = collect_data()

    # Define training data
    X_train = np.array(images)
    y_train = np.array(measurements)

    print('Shape image data: ', X_train.shape)
    print('Shape measurement data: ', y_train.shape)

    return X_train,y_train

def collect_data():
    # init image and measurement buffer
    images = []
    measurements = []

    # open csv file from simulation data
    for counter,data_path in enumerate(data_paths):
        if not data_usage[counter]:
            continue
        print(directory + data_path + "loading..")
        # init csv file buffer
        lines = []

        with open(directory + data_path + 'driving_log.csv') as csvfile:
            reader = csv.reader(csvfile)
            for line in reader:
                lines.append(line)

        # loop through frames and store images and measurements
        for line in lines:
            source_path = line[0]
            file_name = source_path.split('/')[-1]
            current_path = directory + data_path + 'IMG/' + file_name
            image = mpimg.imread(current_path)
            images.append(image)
            measurement = float(line[3])
            measurements.append(measurement)
            if use_multiple_cameras:
                if not line[1] == "":
                    source_path = line[1]
                    file_name = source_path.split('/')[-1]
                    current_path = directory + data_path + 'IMG/' + file_name
                    image = mpimg.imread(current_path)
                    images.append(image)
                    measurement = float(line[3]) + correction_factor
                    measurements.append(measurement)
                if not line[2] == "":
                    source_path = line[2]
                    file_name = source_path.split('/')[-1]
                    current_path = directory + data_path + 'IMG/' + file_name
                    image = mpimg.imread(current_path)
                    images.append(image)
                    measurement = float(line[3]) - correction_factor
                    measurements.append(measurement)

    # data augmentation
    if data_augmentation:
        extend_with_flipped_images(images,measurements)

    return images,measurements


def extend_with_flipped_images(images,measurements):

    print("Before data augmentation = " + str(len(images)))
    # determine
    number_of_images = len(images)
    for i in range(number_of_images):
        # flipped data
        images.append(cv2.flip(images[i],1))
        measurements.append(measurements[i]*-1.0)

    print("After data augmentation = " + str(len(images)))

def draw_data(y):
    plt.hist(y, 29, facecolor='g', alpha=0.75)
    plt.title('Histogram of steering angles')
    plt.grid(True)
    plt.show()