This repository consists of all the project assignments for course ENPM673 at University of Maryland. This course is specifically designed to give insights to Robotic perception and includes topics from the very basics like various image transformations to state-of-the-art algorithms like monocular visual odometry. In this section, brief descriptions are provided about every project done under this course. The projects done in this course are listed below, please click on the link to reach a particular section:

• Lane Detection
• Visual Odometry
• Buoy Detection
• Traffic Sign Detection
• Optical Flow
• Countingpins

The above implementation has been carried out on MATLAB 2017a. Also note that for Visual Odometery project you will need Computer Vision toolbox of MATLAB by Mathworks Inc. It can be found here.

Lane Detection: Detected and tracked lanes from the video given. Turn Prediction is also included.

Calculated the trajectory of a camera inside a moving car. The car circles around a block. Comparisons between matlab used Function and User Defined Functions are also made.

Continuously tracked 3 differently colored underwater buoys.

Used MSER features and an SVM classifier to detect traffic signs and make a bounding box around them.

Compared different methods like Lucas-Kanade, Farneback and Horn-Schunck to get optical flow on Middlebury grayscale dataset.

Identify and count the number of pins in the image. (included white and transparent pins as well)