OK, everyone, I'm signing off for today. If you think of more questions, feel free to post them in the chat attached to tomorow's class. I'll be there to answer questions with Eric.
BTW, It took me more like 10 min to register and login for tomorrow's goodie. The registration page did not ask for or assign a password, I had to go thru the lost password process to get one. FYI
Do you have any comments using neural networks for vison applications?
Sorry, no. I haven't used neural networks since grad school, a loooong time algo. However, I know that some companies are using algorithms (and in some cases hardware architecturse) that are inspired by the human visual system. It's an intrigueing idea, considering how powerful the human visual system is.
The Zed board sell for 300$ retail, so I'd think the BOM there would be arond 150$ or so with ZYNQ part no more that 70-80$ ... it's the 2nd smallest part of course but still decent
If my application has to recognize size, color and texture from tomatoes, it suitable to use the EVS from National Instruments wich implements a CPU in a FPGA ??, or can i use a lower cost processor?
Much will depend on rates-- how quickly are things moving, how quickly do you need the result, how many tomatoes are in one frame. Depending on these and other parameters, a CPU may be adequate (perhaps ith GPU assistance), or you might need more oomph.
@dhaneshm I've been playing with Zynq for the last 2 months, and will be happy to share my experence. Send me an email to tema8@stanford.com if you have questiions
@zogoto3000: the issue with firefox and audio IS the latest flash. Downgrading flash a release or 2 (to 11.3.300.27x) resolved the audio issues for me. But now IE has no flash since process starts with uninstalling flash.
Jeff Bier, If my application has to recognize size, color and texture from tomatoes, it suitable to use the EVS from National Instruments wich implements a CPU in a FPGA ??, or can i use a lower cost processor?
Alex, Size and color would not be a problem for a SOC (ARM). I am not sure what you mean by texture.
ONE temptation with mobile processors is the possibility whether you can do any meshed processor arrays, any network-based offload...
Yes, this has potential. Also, it's interesting to think about what can be done by combining image data from multiple mobile devices. Check out this amazing project, for example: http://grail.cs.washington.edu/rome/
@artem...I had seen the board..(ESC desk XILINX) with ARM-9 Cortex processor....I was lookin out for voice processing.. it is one of the really cool stuff..I hav'nt dont much experimentation on it.. want to try it out..
Agreed, the ease of use is a big question. Right now I'm evaluating Zynq with the goal of building a next generation camra for computational photography and computer vision. It'll be the next version of Franken camera, if you are familiar with it: http://graphics.stanford.edu/papers/fcam/
Jeff Bier, If my application has to recognize size, color and texture from tomatoes, it suitable to use the EVS from National Instruments wich implements a CPU in a FPGA ??, or can i use a lower cost processor?
@jeff: amongst the four processor choices you presented, what is the most costly efficient way to start programming embedded vision applications?
That depends on what you mean by cost-efficient. Are you talking about the cost of stuff you have to buy to get started? The effort required to do the development? Or the cost of producing your product using that processor?
Your slide talks about implementing CPU in FPGA, but there is a new class of FPGA with embeded ARM A-9 core which in my mind is one of the best options for embeded vision. ZYNQ from Xilinx and HPS from Altera. Have you looked at those chips?
I'm glad you asked! I wanted to mention Zynq but ran out of time. I thihk the combination of the high-performance CPU subsystem integrated with the FPGA on one chip is very promising for many embedded vision applications. The key challenge will be making it easy to use.
Your slide talks about implementing CPU in FPGA, but there is a new class of FPGA with embeded ARM A-9 core which in my mind is one of the best options for embeded vision. ZYNQ from Xilinx and HPS from Altera. Have you looked at those chips?
Any of the Boston events going to be webcast ? Very interested in the material but I'm not able to travel to Boston.
Sorry, no webcast from the Boston event, but many of the presentations will be available as videos on the Embedded Vision Alliance web site after the event.
The BDTI OpenCV Executable Demo Package is an easy-to-use tool which allows anyone with a Windows computer and a web camera to experiment with some of the algorithms in OpenCV v2.3. After downloading the installer zip file, double-click on the zip file to uncompress its contents, then double-click on the setup.exe file.
FPGAs let you do "generic" modular tools; all the required I/O, memory, multi-purpose re external sensors, etc -- and "roll your own" internals specific to YOUR application requirements.
mobile devices generally have TIGHT restrictions on useage, usually don't have much excess space available -- keeping people from stomping on available memory, etc, can be a problem... App dev space IS pretty interesting -- check out "String", among others re AR video overlays...
@Ann - biggest headache was dealing with real time video stream... it's been quite a while ago - by now these problems are long gone (more powerful h/w, etc...)
Suggestion for @Ann -- I'm finding it would be useful (simplify what I'm doing in the background) if the PDF could include some active links (opening into a new tab/page) with samples of an ASSP for example, or a GPGPU, etc -- hardware references, since I'm online and slightly info-starved while I'm listening to voice (which is great, just slower than I can read)... Too bad that URLs embedded here aren't clickable (unless people use the "html" link under the chat window?)...
No specific "typically use"; yes, have implemented; memory useage, accessing the program data-space...
Vision application: starting with a data stream of not-exactly-image-data and generating connected-surfaces, looking to recognize object position-and-orientation info within the field-of-view of a sensor.
Hey, all -- finally caught up to "present time" on the chat window. Audio is working (great!) and PDF slide deck downloaded. Listening to "refresher/update"...
Mac users: yesterdays' looping issues aside, I (Mac OS 10.7 Lion) more generally was unable to hear either yesterday or Monday's lectures via either up-to-date Safari or Firefox via the up-to-date Flash plugin. However, Google Chrome worked (and today is once again working) fine for me.
I also observed from the comments that Windows folks running Firefox were also having problems. I'd suggest either IE or Chrome in this case; either seemed to work for others.
If you're having audio issues, please note that some companies block live audio streams. If you don't hear any audio, try refreshing your browser. The show will be archived and available on this page.
Looking forward to learning how low you can go in obtaining vision information.Wondering if anything is within reach of microcontrollers, or if it needs a multi-chip solution.
It is definitely possible to do some simple vision processing on a microcontroller. It all depends on your data rate (resolution x frame rate) and the complexity of your algoriths. If you just wanted to do face detection at close range, for example, and could tolerate latency of perhaps one second, that would likely be doable on an MCU. Most vision functions will require more processing power, however. Also, interfeacing image sensors to MCUs can be a challenge.
I also observed from the comments that Windows folks running Firefox were also having problems. I'd suggest either IE or Chrome in this case; either seemed to work for others.
Mac users: yesterdays' looping issues aside, I (Mac OS 10.7 Lion) more generally was unable to hear either yesterday or Monday's lectures via either up-to-date Safari or Firefox via the up-to-date Flash plugin. However, Google Chrome worked fine for me.
Looking forward to learning how low you can go in obtaining vision information.Wondering if anything is within reach of microcontrollers, or if it needs a multi-chip solution.
I apologize again for the problem with the streaming audio during yesterday's class. I hope that by now, everyone has had a chance to listen to the archived stream. However, if you haven't had a chance to do so, it's OK to wait until after today's class -- yesterday's session and today's are independent.
I'm thinking about "time of flight" for vision systems and seems it should use a reference pulse of light to measure the time-of-flight.
Do these systems use an expanded time sampling system to stretch 1ft/ns (2ft/ns RADAR time) to something much more measurable?
There are two methods used to measure TOF. The first method measure the TOF directly using counters on each pixel clocked in the Ghz range. The second method modulates the pulse using RF frequencies and measure TOF by the phase difference between the outgoing and incoming light pulse.
The streaming audio player will appear on this web page when the show starts at 2pm eastern today. Note however that some companies block live audio streams. If when the show starts you don't hear any audio, try refreshing your browser.
As energy efficiency becomes more and more a concern for makers of electronics devices, researchers are coming up with new ways to harvest energy from sound vibration, footsteps, and even electromagnetic fields in the air.
The government wants to study your brain, and DARPA wants to use similar information to give robots true autonomy beyond any artificial intelligence developed to date. Sound like science fiction? It's not.
From Dell / Intel® New Paradigms in Design Work Scott Hamilton, vertical market strategist for Dell Precision workstations, 5/2/2013 3
Early in my career, I worked as a draftsman and remember the days of drawing on vellum with numbered pencils and Mylar with plastic lead. This was a fun experience in the sense that I ...
I've been using workstations for more than 10 years and love finding ways to get more performance from my system. With demanding professional applications that require more power each ...
A lasting memory from my first job as an engineer in an auto assembly plant is standing on hard concrete at six in the morning, vending-machine coffee clutched in hand, listening to ...
A quick look into the merger of two powerhouse 3D printing OEMs and the new leader in rapid prototyping solutions, Stratasys. The industrial revolution is now led by 3D printing and engineers are given the opportunity to fully maximize their design capabilities, reduce their time-to-market and functionally test prototypes cheaper, faster and easier. Bruce Bradshaw, Director of Marketing in North America, will explore the large product offering and variety of materials that will help CAD designers articulate their product design with actual, physical prototypes. This broadcast will dive deep into technical information including application specific stories from real world customers and their experiences with 3D printing. 3D Printing is
To save this item to your list of favorite Design News content so you can find it later in your Profile page, click the "Save It" button next to the item.
If you found this interesting or useful, please use the links to the services below to share it with other readers. You will need a free account with each service to share an item via that service.
Tweet This
7 
