I was looking for a computer vision library for D. I know opencvd package exists. It uses OpenCV C interface.

I also stumbled upon DCV. It is a part of mir meta package. It seems it is written from scratch. I really liked the code examples. It stays true to the D spirit, so to speak. So I decided to revive it on my free time.

I'd like to get some help if possible and/or any advise that will help me save time and prevent me from going astray on my quest.

Who is the maintainer of DCV?

I was fixing all the 'depreciated' features and errors, and came across this issue:

.dub/packages/mir-algorithm-3.10.91/mir-algorithm/source/mir/ndslice/internal.d(192,5): Error: static assert:  "dimension = 4LU at position 2LU should be less than N = 4LU
- - -
Error in function
mir.ndslice.dynamic.transposed!(0LU, 1LU, 4LU).transposed!(SliceIterator!(ZipIterator!(float*, float*), 1LU, mir_slice_kind.contiguous), 4LU, mir_slice_kind.universal).transposed
- - -
Function prototype
mir.ndslice.dynamic.transposed!(0LU, 1LU, 4LU).transposed!(SliceIterator!(ZipIterator!(float*, float*), 1LU, mir_slice_kind.contiguous), 4LU, mir_slice_kind.universal).transposed(Slice!(SliceIterator!(ZipIterator!(float*, float*), 1LU, mir_slice_kind.contiguous), 4LU, mir_slice_kind.universal) _slice)
_____"
source/dcv/multiview/stereo/matching.d(240,41):        instantiated from here: `transposed!(SliceIterator!(ZipIterator!(float*, float*), 1LU, mir_slice_kind.contiguous), 4LU, mir_slice_kind.universal)`
source/dcv/multiview/stereo/matching.d(188,12):        instantiated from here: `windowCost!((l, r) => reduce!sad(CostType(0), l, r))`
/Library/D/dmd/bin/dmd failed with exit code 1.

transposed comes from mir-algorithm.

I understand the constraints of transposed. It checks DimensionCTError. But I'm having difficulties understanding what DCV code does. It would be great if someone could help me with this.

for(size_t d = 0; d < props.disparityRange; d++) {
    costVol[0 .. $, 0 .. d, d] = CostType.max;
    import mir.ndslice.dynamic;
    costVol[0 .. $, d .. $, d] = zip!true(lpad[0 .. $, d .. $], rpad[0 .. $, 0 .. $ - d])
    .pack!1
    .windows(windowSize, windowSize)
    .unpack
    .universal
    .transposed!(0, 1, 4) // IT SEEMS ERROR OCCURS AT THIS LINE (?)
    .pack!2
    .map!(x => fun(x.unzip!'a', x.unzip!'b'))
    .pack!1
    .map!sum
    .unpack;
}

Any help is greatly appreciated.

On Friday, 29 October 2021 at 03:47:37 UTC, Ki Rill wrote:

Before creating the binding (opencvd) I asked the same question by opening an issue for DCV. One of the maintainers said it is dead. It would be nice if some people will revive it. I don't have much time nowadays. But i would contribute more to this in the future.

On Friday, 29 October 2021 at 03:47:37 UTC, Ki Rill wrote:

You could maybe try contacting Henry Gouk for more information

https://henrygouk.com/

On Friday, 29 October 2021 at 03:47:37 UTC, Ki Rill wrote:

for(size_t d = 0; d < props.disparityRange; d++) {
    costVol[0 .. $, 0 .. d, d] = CostType.max;
    import mir.ndslice.dynamic;
    costVol[0 .. $, d .. $, d] = zip!true(lpad[0 .. $, d .. $], rpad[0 .. $, 0 .. $ - d])
    .pack!1
    .windows(windowSize, windowSize)
    .unpack
    .universal
    .transposed!(0, 1, 4) // IT SEEMS ERROR OCCURS AT THIS LINE (?)
    .pack!2
    .map!(x => fun(x.unzip!'a', x.unzip!'b'))
    .pack!1
    .map!sum
    .unpack;
}

I also stuck there :) Here is my dcv revision. I believe everything works now, except the stereo module. Maybe, we share our experiences to revive dcv. I don't know; perhaps, you have already solved the problems?

https://github.com/aferust/dcvrev