This is my Honors Project for UMass Lowell fall 2017 semester. It is a software application available on the web which is used to analyze the two-dimensional motion of objects. Friendly to students young and old, it consists of an interface allowing the user to upload a video and then identify an object in the frame. From there, computer vision is used to track this object's motion throughout the video. After a short wait, the user can view the position, velocity, and acceleration of the object with respect to time, automatically calculated for them and placed in a table. There is also the option to see video playback with the computer's calculated coordinates marked on each frame. As a final step, the data can be uploaded to iSENSE, a web page used for storing and visualizing data, enabling effortless plotting of the motion.
Project URL: bit.ly/physics-track
I'm working on some other stuff and wanted to save money on AWS, so Physics Tracker is currently shut down. Feel free to contact me from the email on my Github page if you're interested in using Physics Tracker.
Check out some results on iSENSE
If installing this app on your own server, please install the following dependencies. I have tested on Ubuntu 16.04 machines.
sudo apt update
sudo apt install rails
sudo apt install python
sudo apt install libopencv-dev python-opencv
Sudo apt install ffmpeg
sudo apt install x264
bundle install
rails server
This project will produce a piece of software used to analyze two-dimensional motion of objects. Friendly to students young and old, the software will consist of an interface allowing the user to upload or record a video and then identify an object in the video. From there, computer vision will be used to track the object’s motion throughout the video. The user will then be able to view the position, velocity, and acceleration of the object with respect to time, automatically calculated for them and placed in a table. They will also be able to watch the video again, frame-by-frame, showing the data values at any particular time. As a final step, the data can be uploaded to iSENSE, a web page used for storing and visualizing data. This will enable students to effortlessly graph the data after they collect it.
This software will run on the web, with a mobile app developed subsequently.
As a second part to the project, the resulting application will be demonstrated to teachers and be tested with middle school students from Methuen, MA in collaboration with Fred Martin, and possibly more students of various ages in collaboration with the Physics Department. The teachers will provide feedback about how the application was used in their classrooms over a period of two weeks. So, in addition to the success of the application on its own, its use as a teaching tool will also be evaluated.
While software of this nature certainly exists, the purpose of this particular project is to produce software which is free, accessible, and child-friendly. However, in many ways it will fulfill the needs of video experiments held in some first-year physics courses at UMass Lowell, so its use isn’t restricted to young children by any means. Furthermore, the ability to record your own video rather than using one provided by an instructor will certainly make the learning experience more immersive.
Computer vision has limitations which means subjects of the videos will require clearly-defined edges or colors that contrast to be detected. The goal of this project is not to develop better computer vision algorithms than those available today, but rather to focus on experience of using computer vision for educational purposes. Therefore, the users will simply be instructed to use bright objects or bright lighting when they record their video.
The computer vision will be achieved using the OpenCV library with the Python programming language. An API to this code will be set up on a web server which will most likely be hosted on an Amazon Web Services virtual machine. A web interface will be designed, and later an Android app which communicates with the server. Finally, the iSENSE API will be used to send the data to iSENSE.
As mentioned, the software will run on a web server, and the final products will be both a web page and a mobile application which interact with the web server to perform the video analysis. There will also be a write-up discussing two main topics. First, the degree of success of the application itself, including how limited by lighting and colors the video analysis is, as well as how much the outputted values agree with what would be predicted by kinematics. The discussion will also include reflections on showing the application to teachers and students.
The presentation of this project will entail a live demo on a computer followed by a discussion of the results found from sharing the software with teachers and students.