Self driving car keras dave 2 CNN
CNNs is widely using to build self-driving cars. Nvidia's Dave2, the model that using for autonomous cars, one of the easiest ways to build a Self-driving car. In this project, we implement Dave-2 to Keras and predict steering angle.
Dave-2 model using road images as inputs and predict the vehicle controls (steering angle) as outputs. You can find the full paper HERE
Dataset consists of 45,406 images that get from a camcorder in a real car ride. It was made by Sully Chen. You can download HERE. If you want to train your own model rather than using pretrained, just put the images to a data/ folder.
You can either use pretrained model or train your own.
Note: Because of the largeness of the dataset, you may need high compute resources for training.
To Train:
python3 train.py
To run test set:
python3 run.py
-https://arxiv.org/pdf/1604.07316.pdf