OpenVX 1.2 sample implementation doesn't build with Visual Studio 2017.
Did anyone try to build it for Visual Studio?
As one can see from the output below, errors refer to

• doing pointer arithmetic with void*
• using arrays of variable size.

Severity Code Description Project File Line Suppression State
Error C2133 'ptr': unknown size openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_copy.c 401
Error C2133 'space_weight': unknown size openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_bilateral_filter.c 34
Error C2133 'space_weight': unknown size openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_bilateral_filter.c 94
Error C2036 'void *': unknown size openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_bilateral_filter.c 67
Error C2036 'void *': unknown size openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_bilateral_filter.c 68
Error C2036 'void *': unknown size openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_bilateral_filter.c 75
Error C2036 'void *': unknown size openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_bilateral_filter.c 127
Error C2036 'void *': unknown size openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_bilateral_filter.c 128
Error C2036 'void *': unknown size openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_bilateral_filter.c 135
Error C2036 'void *': unknown size openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_matchtemplate.c 287
Error C2036 'void *': unknown size openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_matchtemplate.c 299
Error C2036 'void *': unknown size openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_tensor_lut.c 34
Error C2036 'void *': unknown size openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_tensor_lut.c 35
Error C2036 'void *': unknown size openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_tensor_lut.c 45
Error C2036 'void *': unknown size openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_tensor_lut.c 46
Error C2036 'void *': unknown size openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_tensor_transpose.c 56
Error C2466 cannot allocate an array of constant size 0 openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_bilateral_filter.c 34
Error C2466 cannot allocate an array of constant size 0 openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_bilateral_filter.c 94
Error C2466 cannot allocate an array of constant size 0 openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_copy.c 401
Error LNK1181 cannot open input file '......\kernels\c_model\Debug\openvx-c_model-lib.lib' openvx-c_model C:\openvx_sampleBIN\sample\targets\c_model\LINK 1
Error C2057 expected constant expression openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_bilateral_filter.c 34
Error C2057 expected constant expression openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_bilateral_filter.c 94
Error C2057 expected constant expression openvx-c_model-lib C:\openvx_sample\kernels\c_model\c_copy.c 401

Hi, For documentation purpose and project release process we need the Export control classification number(ECCN) for openvx if it applies. Can you please let me know the ECCN ?

In specification it is said that output will have [output_width, output_height, #IFM, #batches] dimensions. But, as we could see in samples, it must have dimensions: [output_width, output_height, #IFM, roi_count, #batches].

Problem

Type error of the output tensor.

Documentation says that the type of the output tensor bins of vxHOGCellsNode function is VX_TYPE_INT16.
But there is VX_TYPE_INT8 for bins in CTS.

Links

1. Standard
2. Conformance Tests

1. There is a Katex parsing error on this page (https://www.khronos.org/registry/OpenVX/specs/1.3/html/OpenVX_Specification_1_3.html)

ParseError: KaTeX parse error: Expected '}', got '\root' at position 10: \frac{1}{\̲r̲o̲o̲t̲ 4 \of {2}}

2. On this page in the 1.3 Provisional Specification under 1.1 Purpose, this link is broken: https://www.khronos.org/registry/OpenVX/specs/1.3/OpenVX_Specification_1_3.html

3. Based on issue #38 I realize this might not get fixed, but posting for completeness. The following URLs have missing images:

https://www.khronos.org/registry/OpenVX/specs/1.0.1/html/d5/d61/group__group__log.html
missing -> https://www.khronos.org/registry/OpenVX/specs/1.0.1/html/msc_log_callback.png

https://www.khronos.org/registry/OpenVX/specs/1.0.1/html/da/d83/group__group__user__kernels.html
missing -> https://www.khronos.org/registry/OpenVX/specs/1.0.1/html/msc_cdf_installation.png

https://www.khronos.org/registry/OpenVX/specs/1.0/html/d5/d61/group__group__log.html
missing -> https://www.khronos.org/registry/OpenVX/specs/1.0/html/msc_log_callback.png

https://www.khronos.org/registry/OpenVX/specs/1.0/html/da/d83/group__group__user__kernels.html
missing -> https://www.khronos.org/registry/OpenVX/specs/1.0/html/msc_cdf_installation.png

The definition of the Graph object says the it "Does Not loop back" Does a Accumulate node not loop back to it's input?

Clicking on the link on the README.md page in the OpenVX-Registry (https://cvs.khronos.org/svn/repos/registry/trunk/public/vx/) results in a page with Forbidden access. The permissions in Subversion are set to '* = ', resulting in no one having read access. Suggest we either remove the link or adjust the permissions on the SVN VM.

The following files appear to be missing from the OpenVX Register on khronos.org:

https://www.khronos.org/registry/OpenVX/specs/1.0.1/html/d8/d69/d8/d69/group__group__node__callback.html
https://www.khronos.org/registry/OpenVX/specs/1.0.1/html/df/d09/df/d09/group__group__image.html
https://www.khronos.org/registry/OpenVX/specs/1.0/html/d8/d69/d8/d69/group__group__node__callback.html
https://www.khronos.org/registry/OpenVX/specs/1.0/html/df/d09/df/d09/group__group__image.html
https://www.khronos.org/registry/OpenVX/specs/1.1/html/d4/dae/db/dba/group__group__matrix.html
https://www.khronos.org/registry/OpenVX/specs/1.1/html/d6/d28/d6/d28/group__group__hint.html
https://www.khronos.org/registry/OpenVX/specs/1.1/html/d8/d69/d8/d69/group__group__node__callback.html
https://www.khronos.org/registry/OpenVX/specs/1.1/html/df/d09/df/d09/group__group__image.html
https://www.khronos.org/registry/OpenVX/specs/1.2/html/d0/de3/citelist.html
https://www.khronos.org/registry/OpenVX/specs/1.2/html/d4/dae/db/dba/group__group__matrix.html
https://www.khronos.org/registry/OpenVX/specs/1.2/html/d6/d28/d6/d28/group__group__hint.html
https://www.khronos.org/registry/OpenVX/specs/1.2/html/d8/d69/d8/d69/group__group__node__callback.html
https://www.khronos.org/registry/OpenVX/specs/1.2/html/df/d09/df/d09/group__group__image.html
https://www.khronos.org/registry/OpenVX/specs/1.3/OpenVX_Specification_1_3.html

Problem

1. The input tensor may have 2 or 3 dimensions.
   In case of 3 dimensions the size of the first dimension can be 1 or 2.
   Why it can be changed from 1 to 20 in CTS?
2. According to documentation supported border mode shall be replicate or constant. Why it is set to undefined in CTS? Since it is undefined in CTS, then why the resulting values of borders are still checked?
3. In case of 3 dimensions the first dimension is "radiometric" and the filtering goes by summing averaged weighted pixels in the window.
   Weight for every pixel in the window depends on the distance to the center of the window and on the difference in radiometric space between these two pixels.
   Why is radiometric space ignored in CTS?

Links

1. Standard
2. Conformance Tests

@rgiduthuri @kiritigowda and other interested parties: heads-up that I propose to rename the default branch of this repository to 'main' on or after 2021-08-09, following emerging Khronos & github practice. This will have very little impact, although if you have a checked-out repository clone, you may want to follow the simple instructions github will pop up on the repository webpage describing how to do the same locally.

If there is a concern about this, please raise it here by 2021-08-06.

Problem

1. Why the third features' dimension aren't the same in CTS and in documentation? (therefore, the further formation of the output tensor is incorrect)

   e.g. in CTS for the size of the third features' dimension it's not enough multiplication by (cells_per_block_w * cells_per_block_h)

2. In CTS TEST(HogFeatures, testNodeCreation): why bins_dims is passed instead of features' dimensions?

Links

1. Standard
2. Conformance Tests

I'm sorry to open an issue about this, but I'm not able to extract the tar archive compressed in the file openvx-extension-headers-1.2.tar.bz2.

I'm using 7zip on Windows, and I'm able to extract all other compressed archives.

Hi,
We downloaded the latest available openVx sample code (openvx_sample_1.2.tar.bz2), built the sample libraries and ran the generated unit test executable as listed in the README file (..\openvx_sample_1.2\openvx_sample\README) . Note that our environment is Linux. On running the executable, the log is as follows :

[VX_TEST] Kernels:64 Modules:1 Refs:92
[PASSED][00] Framework: Load XYZ Extension
[PASSED][01] Framework: Load Kernel Node
[FAILED][02] Framework: Copy, error = -12
[FAILED][03] Framework: Copy Virtual, error = -12
[FAILED][04] Framework: Heads, error = -12
[FAILED][05] Framework: Unvisited, error = -12
[FAILED][06] Framework: Virtual Image, error = -12
[vx_test_framework_delay_graph:876] Passed!
[FAILED][07] Framework: Delay, error = -12
[PASSED][08] Framework: Kernels
[FAILED][09] Direct: Copy Image, error = -12
[FAILED][10] Direct: Copy External Image, error = -12
[FAILED][11] Graph: ColorBars YUV, error = -12
Found 0 differences between images (status=0)
[FAILED][12] Graph: ColorBars RGB, error = -12
Min Value in AbsDiff = 0, at 285,40
Max Value in AbsDiff = 81, at 32,32
AbsDiff Mean = 43.941765
AbsDiff Stddev = 23.627544
histogram[0] = 2623
histogram[1] = 2589
histogram[2] = 2235
histogram[3] = 2371
histogram[4] = 2805
histogram[5] = 2685
histogram[6] = 2849
histogram[7] = 2938
histogram[8] = 2988
histogram[9] = 4863
histogram[10] = 4142
histogram[11] = 3359
histogram[12] = 3744
histogram[13] = 3776
histogram[14] = 3701
histogram[15] = 12332
[FAILED][13] Graph: bikegray, error = -12
Nodes[0] average exec: 0.034ms
Nodes[1] average exec: 14.983ms
Nodes[2] average exec: 1.871ms
Nodes[3] average exec: 3.625ms
Nodes[4] average exec: 4.208ms
Nodes[5] average exec: 3.390ms
Nodes[6] average exec: 1.457ms
Nodes[7] average exec: 2.159ms
Nodes[8] average exec: 12.929ms
Nodes[9] average exec: 0.798ms
Nodes[10] average exec: 1.702ms
Nodes[11] average exec: 0.064ms
Nodes[12] average exec: 0.076ms
Nodes[13] average exec: 0.067ms
Nodes[14] average exec: 0.118ms
Nodes[15] average exec: 0.125ms
Nodes[16] average exec: 0.110ms
Nodes[17] average exec: 0.068ms
Nodes[18] average exec: 0.051ms
Nodes[19] average exec: 0.067ms
Graph average exec: 47.911ms
[FAILED][14] Graph: Lena, error = -12
Multiple writers failed with -23!
[FAILED][15] Graph: Accumulates, error = -12
[FAILED][16] Graph: Bitwise, error = -12
[FAILED][17] Graph: Arithmetic, error = -12
There are 0 number of points in the harris array!
There are 482 number of points in the fast array!
Rectangle from Gaussian is {1, 1, 159, 119}
[FAILED][18] Graph: Corners, error = -12
[PASSED][19] Graph: Tracker
Passed 4 out of 20

                Note in the above failures list (all 16 of them) that the error status is always -12 (i.e VX_ERROR_INVALID_REFERENCE) . Can someone please throw some light on what could be leading to this error for the 16 example blocks? Your help is much appreciated.

Thanks,
Srinivas

Issue in openvx_sample_1.1.tar.bz2
VX_CONTEXT_TARGETS defined at openvx_sample/include/VX/vx_ext_target.h, line 57 as :
VX_CONTEXT_TARGETS = VX_ATTRIBUTE_BASE(VX_ID_KHRONOS, VX_TYPE_CONTEXT) + 0xC

VX_CONTEXT_TARGETS defined at openvx_sample/include/VX/vx_types.h, line 806 as :
VX_CONTEXT_IMMEDIATE_BORDER_POLICY = VX_ATTRIBUTE_BASE(VX_ID_KHRONOS, VX_TYPE_CONTEXT) + 0xC

Fix:
VX_CONTEXT_TARGETS defined at openvx_sample/include/VX/vx_ext_target.h, line 57 should be :
VX_CONTEXT_TARGETS = VX_ATTRIBUTE_BASE(VX_ID_KHRONOS, VX_TYPE_TARGET) + 0xC

Below text should be changed to as "Input typing - Every input edge (Dx,Ny) requires Type(Ny, Name(Dx,Ny)) in {INPUT} or {INOUT}"

https://www.khronos.org/registry/OpenVX/specs/1.2.1/OpenVX_Specification_1_2_1.html#sec_data_virtual
2.8.6. Graph Formalisms
Output typing - Every output edge (Nx,Dy) requires Type(Nx, Name(Nx,Dy)) in {OUTPUT, INOUT}
Input typing - Every input edge (Nx,Dy) requires Type(Ny, Name(Dx,Ny)) in {INPUT} or {INOUT}

Problem:

Error in the dimensions of the output map in CTS(OpenVX Conformance Test Suite).

Quote from standard:

The output of the template matching node is a comparison map. The output comparison map should be the same size as the input image.

Why in CTS the output map has following dimensions:

result_width = source_width - template_width + 1 ,
result_height = source_height - template_height + 1

instead of (source_width, source_height) ?

Links:

1. Standard
2. Conformance Tests

Hi all,
Please guide me in converting a C/ C++ code to OpenVX 1.1 format. I am a newbie to this and any insight regarding this will be helpful.
@rgiduthuri
Thanks,
Padmasree N.

Hello everyone,
I have a question about the size of the laplacian pyramid.
According to the specification of the khronos, the Laplacian level Li is computed as:

Li = Gi - UpSample(Gi+1). (1)

UpSample(I) is computed by injecting even zero rows and columns and then convolves the result with the Gaussian 5x5 filter multiplied by 4.

After upsample, size of image is 2 times the original size. But to create Gi+1 we need to downsample using VX_INTERPOLATION_NEAREST_NEIGHBOR. After downsample, size of downsample image is ceil(size of prev image * scale).

For example, i have an image 9x9:

• Apply gaussian and downsample: ceil(9x9 * 0.5) = 5x5
• Apply laplacian: 5x5 *2 = 10x10.

In fomula 1, Gi has size 9x9, upsample(gi+1) has size 10x10
so how much will Li have? (9x9 or 10x10)