Inspiration I've never liked waiting in lines, and I have a good feeling no one else does... That's why I'm planning to use HackNC as an opportunity to prototype a potentially useful product that allows clients to count the number of people in an area, especially lines.
What it does It uses an SVM-based classifier in order to detect (and count) the number of people in an image based on HOG features, and I graph the count as a time series in order to prototype a visualization for a web app, which would be shown to end users in order to help them determine whether a location is worth waiting in line for.
How I built it I trained a linear SVM classfier on NIST-produced sample images containing people. Then, I used it to count the number of people from a live webcam feed of images.
Challenges I ran into I was not able to attain a good precision due to time constraints.
Accomplishments that I'm proud of I still managed to visualize data in real-time at a reasonable frame-rate on my PC (to be ported to an embedded system).
What I learned I learned a good amount about machine learning.
What's next for People Counter I plan to produce a start-up out of this product soon.