/Lane_Vehicle_Detection

Vehicle and Lane Detection using Hough Transforms and Sliding Window with ROI Proposals & HOG + SVM in C++ and Python

Primary LanguageJupyter Notebook

Vehicle and Lane Detection

CSS 487: Computer Vision Final Project

Taking steps toward autonomous vehicles.

We developed a lane and vehicle detection program by utilizing various computer vision methods. As a summary, we utilized a Hough Transform for lane detection and a trained support vector machine on HOG (histogram of gradient) image representations and sliding window in order to detect vehicles in a ROI (region of interest). Also added a feature to count the number of vehicles showed in each frame based on the bounding boxes.

Our C++ solution requires C++17

demo

Support Vector Machine Training

-- Performing a Test on the SVM --
 -------- Training SVM ---------
x_train size = 17560
y_train size = 17560
x_test size = 200
y_test size = 200
-- Training Complete --
SVM Test Accuracy = 0.965000

Hough Transform for Lane Detection demo

Histogram of Oriented Gradients & SVM for Vehicle Detection demo