Vehicle Detection and Tracking

Object detection and tracking for vehicles onboard camera using either an OpenCV method or the YOLO Darkflow Convolutional neural network (CNN) library. This is project 5 of the Udacity Self-driving car nano degree, Vehicle Detection and Tracking submission. This project uses two different forms of implementation.

1. SVM classifier using HOG and colour bin features.
2. Fast YOLO method using the python wrapper darkflow. The code is available on github.

For a detailed description and walkthrough of the key areas of the code, visit the project website here.

A video of the object detection in action can be seen below.

Project website

This repository has an accompanying project page, contains the theory and details behind the code. It can be found here.

Getting Started

Download and unzip or clone this repository onto your local machine with

$ git clone https://github.com/Heych88/udacity-sdcnd-vehicle-detection-and-tracking.git

Prerequisites

This project requires python 3 and the following libraries installed.

• OpenCv
• Tensorflow
• numpy
  • $ pip3 install --user numpy
• sklearn
  • $ pip3 install --user scikit-learn
• scipy
  • $ pip3 install --user scipy

Installing

This repository only contains the code for the VM classifier using HOG and colour bin features.

Installing Fast-Yolo

To use the Fast YOLO method, download the python wrapper darkflow as outlined in the darkflow README.md. Note: follow the ‘Getting Started’ installation before proceeding.

Once extracted or cloned, copy the folders cfg, cython_utils, dark, net and utils into the location of this repositories directory. Add a folder named bin and store the downloaded Tiny YOLO weight file from the VOC2007+2012 section, downloaded from here into that folder.

Open the project up in your favourite python ide and In the file objectdetection_YOLO.py, uncomment the lines from net.build import TFNet and self.tfnet = TFNet(options) at lines 3 and 13. In the file main.py, set the variable use_yolo = True

If all has been installed correctly, run main.py and the following image should appear.

Running the Code

Navigate to the directory of the repository in a terminal and run main.py.

$ cd <local directory>/udacity-sdcnd-advanced--lane-finding
$ python3 main.py

The following image should appear.

Built With

• darkflow - The YOLO framework used

License

This project is licensed under the GNU General Public License v3.0 - see the LICENSE.md file for details.