Behavioral-Cloning/generate_data.py at master · Pablitinho/Behavioral-Cloning · 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
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
import pickle

import cv2
import csv
import numpy as np
from PIL import Image

def get_image(line,id,img_path):
    source_path = line[id]
    file_name = source_path.split('/')[-1]
    #file_name = source_path.split('\\')[-1]
    current_path = img_path + file_name
    #image2 = cv2.imread(current_path)
    image = np.asarray(Image.open(current_path))
    return image


def trans_image(image, steer, trans_range_x, trans_range_y):

    rows, cols, chan = image.shape
    # Translation
    tr_x = trans_range_x * np.random.uniform() - trans_range_x / 2
    steer_ang = steer + tr_x / trans_range_x * 2 * .2
    tr_y = trans_range_y * np.random.uniform() - trans_range_y / 2
    # tr_y = 0
    Trans_M = np.float32([[1, 0, tr_x], [0, 1, tr_y]])
    image_tr = cv2.warpAffine(image, Trans_M, (cols, rows))

    return image_tr, steer_ang

def augmentation_shift(image, desplacement):
    num_rows, num_cols = image.shape[:2]

    translation_matrix = np.float32([[1, 0, desplacement], [0, 1, 0]])
    img_translation = cv2.warpAffine(image, translation_matrix, (num_cols, num_rows))
    return img_translation

def create_pickle(log_data,img_path):
    samples = []
    with open(log_data) as csvfile:
        reader = csv.reader(csvfile)
        first_time = True
        for line in reader:

            if first_time == True:
                first_time = False
            else:
                samples.append(line)

    from sklearn.model_selection import train_test_split
    train_samples, validation_samples = train_test_split(samples, test_size=0.2)
    # -------------------------------------------------------------------------------------------------------------------
    # Validation generation
    # -------------------------------------------------------------------------------------------------------------------
    images = []
    measurements = []
    cnt = 0
    num_sub_samples = 0
    for line in validation_samples:
        if num_sub_samples < 16:
           num_sub_samples += 1
           image_center = get_image(line, 0, img_path)

           steering_center = float(line[3])
           images.append(image_center)
           measurements.append(steering_center)
        else:
            images = np.array(images)
            measurements = np.array(measurements)
            print("Batch Validation: ", cnt)
            with open('./augmented_data/images_validation_' + str(cnt) + '.pickle', 'wb') as handle:
                 pickle.dump(images, handle, protocol=pickle.HIGHEST_PROTOCOL)
            with open('./augmented_data/stearing_validation_' + str(cnt) + '.pickle', 'wb') as handle:
                 pickle.dump(measurements, handle, protocol=pickle.HIGHEST_PROTOCOL)
            num_sub_samples = 0
            images = []
            measurements = []
            cnt += 1
    # -------------------------------------------------------------------------------------------------------------------
    # Training generation
    # -------------------------------------------------------------------------------------------------------------------    images = []
    measurements = []
    images = []
    cnt = 0
    num_sub_samples = 0
    for line in train_samples:
        if num_sub_samples < 16:

            num_sub_samples += 1
            image_center = get_image(line, 0, img_path)
            image_left = get_image(line, 1, img_path)
            image_right = get_image(line, 2, img_path)

            steering_center = float(line[3])

            correction = 0.25  # this is a parameter to tune
            steering_left = steering_center + correction
            steering_right = steering_center - correction

            images.extend([image_center, image_left, image_right])
            measurements.extend([steering_center, steering_left, steering_right])
            # -------------------------------------------------------
            # Flip image only when steering
            # -------------------------------------------------------
            if abs(steering_center) > 0.13:
               img_flipped = cv2.flip(image_center, 1)
               images.append(img_flipped)
               measurements.append(-steering_center)

            # -------------------------------------------------------
            # Shift to the left
            # -------------------------------------------------------
            trans_range_x = 30
            trans_range_y = 5
            for iter in range(1, 4):
                #img_translation = augmentation_shift(image_center, -iter*2)
                img_translation, steering = trans_image(image_center, steering_center, trans_range_x, trans_range_y)
                images.append(img_translation)
                measurements.append(steering)
                #measurements.append(steering_center - 0.05 * iter)

        else:
            print("Batch: ", cnt)
            images = np.array(images)
            measurements = np.array(measurements)
            with open('./augmented_data/images_train_' + str(cnt) + '.pickle', 'wb') as handle:
                pickle.dump(images, handle, protocol=pickle.HIGHEST_PROTOCOL)
            with open('./augmented_data/stearing_train_' + str(cnt) + '.pickle', 'wb') as handle:
                pickle.dump(measurements, handle, protocol=pickle.HIGHEST_PROTOCOL)
            num_sub_samples = 0
            images = []
            measurements = []
            cnt += 1
    # -------------------------------------------------------------------------------------------------------------------