We originally intended to use Dragonboard’s image processing capabilities to create a people-counting gate that would monitor the people present, therefore possible seats remaining, in an indoor space such as a cafeteria. As we brainstormed for other ideas that could be reached using similar tools, we came across the topic of safety. Although there are emergency alert systems installed on campus, these systems only function when an accident is taking place. Instead, we intended to make progress on actively preventing accidents by providing students with a real time mapping of all people present on campus.

What it does

This current prototype uses image programming and opencv to count the number of people within the range of the camera. This data is then compared to bluetooth signals received from an arduino. The idea is to have student IDs incorporated a bluetooth system that will send signals from the card to dragon boards located around campus. The dragon board will compare the difference between the people campus, which of those people are students or faculty and which of these are strangers. The location of the dragon boards located throughout campus is also meant to help track the location of the people being Thus this will help build a safety system around campus, that will help students feel safe anywhere and anytime on campus. The project is also meant to be expanded into a phone app that will track the people on campus as dots. These dots will be color coated: blue for students, white for faculty, and red for strangers. A stranger is someone who is not supposed to be here, and will officially be classified as someone who does not have a UCSD ID card with them. This will help the campus police quickly respond to emergencies, as well as reduce overall crime on the UCSD campus.

How we built it

  1. Connect Dragonboard to monitor, keyboard, and webcam
  2. Research on OpenCV and integrating existing image processing algorithms
  3. Connect bluetooth module to Arduino Uno and enable bluetooth signal emission
  4. Analyze image processing data alongside bluetooth data
  5. Print analyzed results to inform users of strangers’ presence

Challenges we ran into

  1. Some parts were already checked out by other groups before our assigned time (we were missing some of the initial products we needed): (a) We adjusted our project to use the available devices: switch from GPS to bluetooth

  2. Dragonboard’s initial failure to reboot, forcing a total flash of data, causing major delay in project progress (a)Troubleshoot to get the system running; worked harder afterwards to finish project. Some parts had to be adjusted (took out part about visualization mapping of the people)

  3. The difficulties to record distance via image processing as well as bluetooth (a)Kept the current prototype to a small scale system, and excluded location tracking in this version of the prototype. Plan to increase the number of Dragonboards in an area to increase accuracy in location tracking. (Triangular Tracking)

  4. Arduino Uno’s initial failure to upload (a)We worked around to find a way to give the Dragonboard a bluetooth signal without needing to directly pair to it

  5. The difficulties to train a human detection system in limited time (a)The data we were using had some issues in recognizing people when their back was turned to it + there were issues with recognizing people if they were leaning or sitting.

Accomplishments that we're proud of

  1. A functioning prototype of the the Dragonboard-bluetooth monitoring system
  2. Preventing the Dragonboard from crashing for the second time
  3. We were able to combine various features of an already existing program, which detected facial and body recognition separately, into a more advanced program that used the combined features to create new analysis to help real world problems.

What we learned

  1. Make sure the development environment is crash-safe before investing coding time
  2. Backup all progress through multiple means
  3. Understanding parameters, such as scaleFactors and minNeighbors, of various image processing algorithms

What's next for Safety

  1. Safety team goes to take a nap
  2. Proceed with GPS sensors to enable location transmission from Arduino to Dragonboard, thereby creating a map that will be operable through phones.
  3. Speed up human detection

Built With

Share this project: