During greater part of 2017 and 2018, our team was doing research in Deep Neural Net (DNN) anomaly detection. The problem we were looking to solve was that control centers usually had far too many surveillance video cameras and control panels to monitor and not enough personal to pay attention to them all. Why not apply machine-learning (ML) anomaly detection on camera and computer monitor feeds to alert control center personal of abnormal events as soon as they occur?
buffering description TBD...
Experiences using this skill are shown below:
This was a student intern project that I took over to turn into a useful application for use within the company. Even though this wasn't a research project, we thought it would be a good way to make Barco Labs better known throughout the company, as many employees viewed us as an "ivory tower" doing esoteric research of little practical value. The Smart Meeting Room App (SMRA) had the very practical benefit of finding and scheduling meeting rooms on the Barco campus.
An earlier version of our machine-learning person tracking software was too slow to keep up on every video frame. (This was before our team attempted to use GPU acceleration with more efficient person tracking software.) This resulted in jumpy video transitions while tracking someone. I was given the task of finding a way of applying video motion smoothing so the resulting video framing would be smooth and professional looking.
I was given the task of researching automatic white-balance algorithms with the goal of calibrating multiple video cameras to the same color balance. The problem was that when switching between multiple cameras covering the same scene, a noticeable color shift was observed in the video stream, especially when the cameras were of different manufacture.
Engineered the high-performance tile-map server technology for the company that provided all the custom map overlays displayed by the universal map applications (above). Features included:
Designed and implemented complex, real-time, universal map display applications in Adobe Flash for the web and CocoaTouch for the Apple iPad that yielded a significant increase in revenue and helped achieve financial independence for the company (according to the President/CEO, Maurice Bailey).
Designed and created all of the Universal Flash Viewer plug-ins including all the graphics art work. They were all implemented in pure ActionScript 3 without the use of the main timeline. All ActionScript code was contained in separate source files. All plug-ins followed a standard design pattern:
Refactored the previous weather plug-in I implemented to do on-demand loading of weather only when requested. The previous approach preloaded the weather in the background to speed delivery but that approach negatively affected overall Viewer performance. The weather animation in the new approach would start up soon as 2 layers were loaded and would automatically add more layers to the animation sequence as each layer became available. This provided on-demand loading with a fast initial presentation.
The redesigned map plug-in solved a long-standing problem with the old map plug-in that was actually a port of some old code written by someone else. The new plug-in required a lot less memory, was a lot less buggy and it could show dozens of different base maps from multiple map tile servers. All of this could be configured without touching any code (unlike route-me and OpenLayers). The map type selected in this screenshot was the USGS topo map. The pushpin labels were draggable and individually displayable.
FlyteComm wanted to put the map tile servers under that same RSA SecurID gateway as the ASP .NET servers but the company's server infrastructure did not allow that plus there were known scalability issues. To demonstrate to IT there was a better way, I setup a VMWare virtualized system to prototype a better server infrastructure. The system consisted of:
Developed a prototype version of the FlyteComm Universal Viewer for the iPhone/iPad as a marketing tool to guage customer interest. This prototype iOS app generated so much customer enthusiasm that the company went full speed ahead developing it into a real product. The iPad/iPhone application was written in Apple's Objective C for iOS 5 in XCode 4.3. There were many challenges. We could not use Apple's MapKit, based on Google Maps, because of Google's licensing restrictions on commercial use.