This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

DaVinci vs Omap application Advise

Other Parts Discussed in Thread: OMAP3530

Hi,

would like some advise regarding an application plataform. Please feel free to delete this post if it is the wrong place, has i havent found one i could fit it into.

Situation:

I pretend to develop smart-cameras, so basically i will perform computer vision algorithms. One application will be traffic control (detecting cars, flow measures, etc), others could be counting persons, use stereo etc...

Looking at the available choice i would naturally go to the DaVinci plataforms, but the Omap ones keep poping up, speacially OMAP 3530.

The difference between this two is not at least to my eyes pretty clear, even TI sometimes labels some common applications to both.

This way, i would like an experienced advise from for what would suits better for my application, since i pretend to buy a developmant kit this month.

Best Regards
 

  • Both platforms can be good for this sort of thing, so it depends more on your specific processing and i/o requirements as well as what your general background is in embedded processing (i.e. are you an ARM/Linux engineer, or a DSP/RTOS engineer, etc.).

    If you want out of the box video I/O for prototyping than one of the Davinci platforms is probably the most suitable as most will have out of the box support for at a minimum a composite video input and output so you can use an off the shelf camera. With the OMAP3 you can do this as well, however it takes additional hardware today, namely the Mistral multimedia daughterboard.

    If you prefer ARM/Linux development and don't want to deal with the DSP than the OMAP is probably your best bet, as it has by far the most powerful ARM with its 600Mhz Cortex-A8, so there is the potential of running many algorithms without involving the DSP at all. On the other hand if you are more interested in DSP/RTOS type development than one of the Davinci devices would be more appropriate, with the far end of the spectrum being the DM6437 as a DSP only device.

    In general if you could provide more specifics on exactly what you want to do, what your video sources and outputs need to be, as well as what kind of processing performance you will need, and the development environment you prefer is, than I could make a more concrete recommnedation.

  • Hi,

    thanks for your quick reply.

    At first i would say i do pretend to run several computer vision algorithms in plataform, for several applications. Still the immediate aplication will be traffic detection. So it is like an inteligent surveillance system, has it may be applied to detect people has well.

    So basiccly i will have 1 camera has input,  will run some algorithms , and then the output will probably need to be sent throw the web.

    Other situations may require two cameras maximum.

    Relating to the algorithm, part of it is implemented by texas in this report, take a look : URL

    Has for the Arm / Dsp developer i would not really want to choose a plataform based on that.

    Other possible applications i should need is: Stereo Vision, and Optical Coherence Tomography(medical one)

    Thanks for your time...

    P.S. Feel free to delete the other topic in OMAP section, i posted in the two of them cause i was expecting that in the DaVinci one, Davinci developers would enphase this ttechnology and in the OMAP they would suggest the OMAP. I though i would need to measure both and get some more ideas about their differences.

    Best Regards

     

     

  • This is an interesting topic. It makes me think that OMAP3 may be a good option for applications
    which is using Davinci currently.
    But I don’t understand that additional hardware is required for OMAP3. I take a look at the document
    of OMAP3530. It doesn’t have a VPSS module but has camera image signal processing module and display module.
    The highest supported frequency of DSP part on OMAP3 is lower than its counterpart on Davinci.
    It is a consideration when decision is made. Will be new OMAP3 with higher DSP frequency introduced?

    Best Regards

    Jogging

  • Filipe Alves said:
    Relating to the algorithm, part of it is implemented by texas in this report, take a look : URL

    The application note you point to is DSP based, so it would apply more to a Davinci device than an OMAP, not that the OMAP could not also do this, but in general you get more DSP performance on a Davinci and more ARM performance on an OMAP, so the Davinci seems more appropriate in this case.

    Jogging Song said:
    The highest supported frequency of DSP part on OMAP3 is lower than its counterpart on Davinci.
    It is a consideration when decision is made. Will be new OMAP3 with higher DSP frequency introduced?

    This is true, and is why in general the DSP heavy designs are more well suited to Davinci, but on the other hand the OMAP has a much more powerful ARM core, so for general purpose processing (i.e. GPP/ARM) heavy designs the OMAP is more suitable. There are of course higher performing products on the roadmap, for an OMAP3 in particular the DSP will probably always be slower than the ARM, though there may be future devices that work differently.