What’s the Project about?
This study will develop methods to detect a person falling, either because of tripping over, fainting, having a heart attack, a stroke or any other event. The outcome of this study would be used to develop software to alert and promote assisted living. This would be of a particular benefit to the elderly population, people who are prone to accidents and people with certain medical conditions such as low blood pressure, vertigo, heart conditions etc. that may result in a fall.
During this study, which will last for one year, I aim to publish my work on academic journals, providing publicly the methodologies and results. There is not any financial benefit for myself or Kingston University for doing this research.
Some historical data of my research
I have studied computer science and physics and I have worked in the area of computer vision for a number of years. Some years ago (Jan 2011), I had the idea to work on a research project regarding fall detection. I was able to take a visiting position at Kingston University London for 6 months to work on this project. For the purposes of the research I used a new at the time device which could perform infrared data recording. The device was Kinect, a popular and well known Xbox game console. I was in fact one of the first machine vision scientists that used Kinect for research purposes other than gaming. Using Kinect, I could process data and with the use of my algorithms, the software was successful in identifying a particular type of fall (fall while walking). Please see the video (https://goo.gl/ZKW1GP) for the demos. However, this research was restricted in its findings and the type of fall it could identify.
Fall detection in action
At the end of the visiting period I published a journal paper (https://goo.gl/mzIx5u) which - based on Google Scholar - has more than 80 citations. The paper was on print by Springer-Verlac on 2014. Also, my work was presented on a Microsoft organised event (Kinect for Life, Oct 2012) to promote the uses of Kinect on other applications.
Phases of the process
Where will your money go?
Recently and after evaluating different Kinect devices and processing software, I discovered that my current hardware is not suitable for running those experiments. On the other hand, the University supplies hardware from a specific supplier, which unfortunately has limited flexibility in building new - custom machines. Therefore, a new laptop is needed urgently to cover the needs. A computer vision suitable laptop costs average on £2,100. To prepare a demonstrable system, I will need 2 Kinect devices on which I will remove the case and motorised tilt and build a nice and smaller white enclosure with a bracket in order to fit it on a wall. Each Kinect costs approximately £100. The bracket and enclosure will cost around £25 for each Kinect. The University will not allow me to break their Kinects and as you understand, those costs need to come from somewhere unfortunately.
To summarise the costs:
1x Laptop £2,100
2x Kinects £200
2x Enclosures with brackets £50
Thank you for reading and I would be grateful for any contributions to this cause.
George Mastorakis Visiting Researcher, SEC, Kingston University London
George counts almost a decade in Image Processing and Machine Vision that has involved image understanding, pattern recognition and algorithmic development, having worked with established Universities and private companies. He is now a Visiting Researcher at Kingston University London, working on accident detection.
The experience and increased interest in several scientific areas (i.e. Engineering, Computer Science and Physics) has allowed him to acquire broad knowledge and abilities to apply combined and innovative thinking but also problem solving strategies using transferable skills from those scientific disciplines.
1. Mentioning of your name on the published research papers as a supporter (£10 and up)
2. Invitation to meet face to face, discuss the progress of research and if possible to attend the experiments live! (£100 and up)
3. Full algorithm (as a pseudocode) will be provided for the most generous contributions of £1,000 or more (the algorithm will be provided after publishing relevant papers in compliance with the intellectual property rights)