Search:

Type: Posts; User: emrainey

Page 1 of 2 1 2

Search: Search took 0.00 seconds.

  1. I don't think there is a "Stitch" kernel by...

    I don't think there is a "Stitch" kernel by itself (you're always welcome to write a user mode kernel to do this and share it here!). There are remaps and ROIs of images, which would have to be...
  2. You may have found a bug in the implementation....

    You may have found a bug in the implementation. You should not have to release the graph and recreate it to change the input image data, or the the input image (it may require re-verification, I...
  3. Are you asking how to access an image? You can...

    Are you asking how to access an image? You can use vxAccess/vxCommit in between graph executions to update the input image data. If you have wrapped some external memory with vxCreateImageFromHandle,...
  4. Replies
    2
    Views
    491

    vxAccessImagePatch is there to guarantee...

    vxAccessImagePatch is there to guarantee consistent access behavior across a broad range of platforms. The example in the API docs for this function show the various "faster" ways to do it, but all...
  5. They are intended to be part of the spec so that...

    They are intended to be part of the spec so that the _results_ of the operations can be rigorously defined and checked by the conformance test. Otherwise you have the same problem as OpenCV. If 2...
  6. Replies
    2
    Views
    2,361

    Go to the Makefile and comment out the lines that...

    Go to the Makefile and comment out the lines that say TARGET_COMBOS. These assign the different builds it can do which may not apply or work on an embedded board.
  7. Replies
    0
    Views
    2,028

    OMX_U32 definition problem

    I've notice that the 1.2 spec still has a problem where OMX_U32 is defined as;


    typedef unsigned long OMX_U32;

    But on most 64 bit platforms this compiles as a 64 bit value! It should be...
  8. Replies
    3
    Views
    1,481

    Merging kernels is an implementor's choice, but...

    Merging kernels is an implementor's choice, but usually (though not always) will be triggered by the user's usage of Virtual Data types.
  9. Replies
    3
    Views
    1,481

    Thank you! Final with respect to 1.0. 1.1,...

    Thank you!



    Final with respect to 1.0. 1.1, 1.2 may expand types. Since we use opaque objects, their internal typing (U8/U16) can be expanded (not replaced) and the API doesn't change....
  10. Eventually OpenVX will be used as the HAL...

    Eventually OpenVX will be used as the HAL (Hardware Acceleration Layer) of OpenCV, so it should be transparent to you. The maintainers of OpenCV will have to publish their dates for having that...
  11. I pulled down the sample package and just did...

    I pulled down the sample package and just did this:



    $ make
    ...
    $ LD_LIBRARY_PATH=out/LINUX/x86_64/release/ out/LINUX/x86_64/release/vx_query


    This worked just fine.
  12. To the point of vx_scalar versus vx_int32's: A...

    To the point of vx_scalar versus vx_int32's:

    A vx_scalar is an opaque data object reference (internally it may be a handle/index/pointer) with no defined (by the specification) internal size...
  13. Replies
    4
    Views
    1,751

    The delay object is a container, not a data...

    The delay object is a container, not a data object like an image. The since this API is C and that does not have overloading, we didn't want to have to have multiple versions of the same Node API,...
  14. Replies
    4
    Views
    1,390

    Possibly

    Any particular architecture/algorithm combination may have different optimization strategies. Sometimes splitting a node across all or some targets approach is faster, sometimes it isn't. This API is...
  15. Replies
    4
    Views
    1,525

    NO it's NOT

    This is NOT an official release of the sample implementation. Whomever did this probably violated several contractual obligations to Khronos and they and their company are probably (though I am not a...
  16. Replies
    4
    Views
    1,751

    One purpose of the delay node is to provide...

    One purpose of the delay node is to provide history to the graph without the user manually reconnecting objects, thus possibly causing the need for graph re-validations between executions. Some algos...
  17. Once the sample implementation is released, it...

    Once the sample implementation is released, it will be more obvious how to do so.

    The really quick howto, in your C code:


    #include <VX/vx.h>
    #include <VX/vxu.h>

    Compile command:
  18. The sample implementation was compilable by the...

    The sample implementation was compilable by the full Android build at one point. I will have to check to see if that is still true. Building the whole implementation via the NDK is also possible, but...
  19. Replies
    5
    Views
    2,264

    It's an attribute of the vx_convolution object. *...

    It's an attribute of the vx_convolution object. *

    vxSetConvolutionAttribute(conv, VX_CONVOLUTION_ATTRIBUTE_SCALE, &scale, sizeof(scale));
  20. Replies
    5
    Views
    2,264

    Also s != 0 - - - Updated - - - Normally s...

    Also s != 0

    - - - Updated - - -

    Normally s is the sum of the coefficients, but it can be set manually too.
  21. Replies
    5
    Views
    2,264

    Scale is the final divisor of the sum. ...

    Scale is the final divisor of the sum.

    example:
    input image = p
    coefficients[3][3] = c
    scale = s
    output value = v

    v = ( (p[0][0] * c[0][0]) + (p[0][1] * c[0][1]) + (p[0][2] * c[0][2])...
  22. Replies
    3
    Views
    2,086

    The Specification API (vxCreateContext,...

    The Specification API (vxCreateContext, vxCreateGraph) are not helpers, they are part of the Specification API. API that initialize a vx_matrix to a specific affine rotation angle, for example would...
  23. Replies
    3
    Views
    2,086

    The sample implementation will be released some...

    The sample implementation will be released some time after the Final Ratification of the specification. The sample will hopefully be a complete and fully working implementation, but it is not...
  24. OpenVX's goal is to accelerate *all* vision...

    OpenVX's goal is to accelerate *all* vision functions, not just image processing. OpenCV has some acceleration mechanisms now, but not all that are possible. OpenVX allows those new possibilities and...
  25. Replies
    1
    Views
    1,358

    3x3 is a very common filter size in real-time and...

    3x3 is a very common filter size in real-time and embedded applications. Some kernels do have internal filters of different sizes. Future support for other sizes is up for debate and we welcome...
Results 1 to 25 of 37
Page 1 of 2 1 2
Proudly hosted by Digital Ocean