Smart Cam - Prof. Sam Siewert
Prof. Sam Siewert came to UAA from University of Colorado Boulder in the fall of 2012 with an Intel-funded research project in progress. Dubbed the "smart cam," its goal is to develop a smaller, faster processing interface for visible and near infrared cameras.Rather than outputting data that requires interpretation and analysis by software or a researcher, the smart cam will have a card or board embedded in it that interprets what the camera sees and relates a scene analysis to the researcher.The long-term fundamental goal for this research is to make progress on the vision to develop computers capable of real-time inverse rendering—in other words, a computer capable of seeing and describing scenes like humans do.In the future, a computer that can see a scene like the human eye does and make sense of it has many applications for automation and autonomous system operations.
The smart cam will be a significant step toward that, leveraging advanced, highly concurrent processing hardware like graphics co-processors, field-programmable gate arrays and multi-core processing, with many practical embedded real-time applications.For example, the USGS has interest in the possibility of using UAV imagery to address the damage done to trees in the Western United States by pine bark beetles. The smart cam described above could be embedded in an unmanned aerial vehicle (UAV), combining Prof. Siewert's knowledge of UAVs and embedded systems with computer and machine vision.As the UAV flies, the camera will be able to differentiate living trees from dead ones.Then, it can record global positioning data (latitude, longitude, and altitude), adjusted based upon spatial awareness, and record accurate coordinates of each dead tree in a compiled a list.The value of this type of application is simple – forestry workers can use the list to locate and remove the dead trees.The camera could also be used to monitor many other natural resources of interest to agencies like the USGS; for example, counting animals and providing efficient population estimates.
At the moment, the project is being worked on at CU Boulder, but Prof. Siewert hopes that with additional funding, he can hire UAA students and set up parallel teams to work together.Prof. Siewert already has plans to encourage collaboration, not just with his students, but with the community.In order to foster creativity and innovation, Prof. Siewert is planning to make the hardware development open source.
Collaborating with other universities is just one of the things Prof. Siewert hopes to do at UAA.UAA's mission and unique location makes healthcare a natural focus for future research in addition to environmental sensing.Prof. Siewert is looking forward to testing the smart cam architecture and machine vision capabilities in Alaska's unique and challenging environment.
Prof. Siewert has already wrapped up several other projects.Among those, he worked on the Spitzer Space Telescope, which recently completed its deep spaced mission. While working on the deep infrared instrumentation embedded systems, he learned to really appreciate the value of instrumentation. He has also just completed work on a project to determine true latency of video taken from a UAV transported to people on the ground over digital links by comparing this to direct line of sight and cabled analog capture and machine vision analysis.It will be a long time before computers can "see" like humans do, but Prof. Sam Siewert is working on advancements to help make it happen.
In April 2013, the Altera University Program donated equipment to Prof. Siewert's Smart Cam research project.
Loading slide show...