Patrick is currently an undergraduate at New Hampshire Technical Institute in Concord, New Hampshire, pursuing a degree in computer science and entrepreneurship. His deep-rooted passion for programming has developed into a drive for knowledge pertaining to almost anything technical. This drive and his artistic creativity have proven to be a very effective combination when approaching problems and creating interesting and effective solutions. Patrick has volunteered for various instructor opportunities in many innovative fields, including 3D printing technologies. Robotics, machine learning, computer vision, web and mobile development and renewable resources are among Patrick’s many potential career interests.
Project: Lecture Attendance Numbers
Goal of the project:
The goal of my project is to count students in large lecture halls so that professors can get a sense of attendance and also gather statistics about their class. This project will continue to be worked on outside of the Summer School and eventually be adapted to detecting parking spaces in parking lots.
Summary of work:
I approached this problem through three different avenues. The first two, one using a neural net and the other relying on a few of Mathematica’s built-in functions, were horribly inaccurate. The third, however, is a much more thought-out process and will likely evolve into the solution I am looking for. This solution is driven by a chunk of image processing paired with, eventually, a classifier. The main goal was to manipulate the Mathematica function ImageDifferences and machine learning to work in a way that catered toward the needs of my project. This image manipulation can be seen inside the project notebook. The image processing worked very well and will only strengthen with the help of machine learning. Due to the multiple attempts and time spent on the first two unrealistic solutions, I did not have the time I would have needed to create an accurate classifier. That said, I was able to create an excellent platform for continued work on this project and I believe that I am on the brink of a solution.
Results and future work:
Overall I would give my project a rating of: not a complete disaster. While I did fail at producing a working product, I have given myself a strong base with which I can continue my work. My other two solutions, while failing, did give me some valuable information and possibly some strategies that I will use later on in my project. Given more time and a refinement on algorithms, I believe that this project has many potential uses outside of lecture halls.
Some features that I am planning to add after the Summer School:
- Class engagement rate (determining when students are paying attention)
- Cellphone use
- True attendance (learning specific students’ appearance and taking a more detailed attendance)