callmemaze

import numpy as np
import pandas as pd
import os
from matplotlib import pyplot as plt
from sklearn.utils import shuffle
from sklearn.model_selection import train_test_split
from imgaug import augmenters as aug
import matplotlib.image as mpimg
import cv2
import random

callmemaze

def balanceBiased(data, display=True):
  nBins = 31
  samplesPerBin = 1000
  hist, bins = np.histogram(data['Steering'],nBins)
  if display:
    center = (bins[:-1] + bins[1:])*0.5
    plt.bar(center, hist, width = 0.06)
    plt.plot((-1,1), (samplesPerBin, samplesPerBin))
    plt.show()
  removeIndex = []

callmemaze

def augmentImage(imgPath, steering):
  img = mpimg.imread(imgPath)
  if np.random.rand() < 0.5:
    pan = aug.Affine(translate_percent={'x': (-0.1,0.1),'y': (-0.1,0.1)})
    img = pan.augment_image(img)
  if np.random.rand() < 0.5:
    zoom = aug.Affine(scale=(1,1.2))
    img = zoom.augment_image(img)
  if np.random.rand() < 0.5:
    brightness = aug.Multiply((0.4,1.2))

callmemaze

def preProcessing(img):
  img = img[60:135,:,:]
  img = cv2.cvtColor(img, cv2.COLOR_RGB2YUV)
  img = cv2.GaussianBlur(img, (3,3),0)
  img = cv2.resize(img, (200,68))
  img = img / 255
  return img

callmemaze

model = Sequential()
model.add(Conv2D(24,(5,5),(2,2), input_shape=(66,200,3), activation="elu"))
model.add(BatchNormalization())
model.add(Conv2D(36,(5,5),(2,2) ,activation="elu"))
model.add(BatchNormalization())
model.add(Conv2D(48,(5,5), activation="elu"))
model.add(BatchNormalization())
model.add(Conv2D(64,(3,3), activation="elu"))
model.add(BatchNormalization())
model.add(Conv2D(64,(3,3), activation="elu"))

callmemaze

callmemaze

callmemaze

callmemaze

callmemaze