<html xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40"><head><meta http-equiv=Content-Type content="text/html; charset=utf-8"><meta name=Generator content="Microsoft Word 15 (filtered medium)"><style><!--
/* Font Definitions */
@font-face
{font-family:"Cambria Math";
panose-1:2 4 5 3 5 4 6 3 2 4;}
@font-face
{font-family:Calibri;
panose-1:2 15 5 2 2 2 4 3 2 4;}
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{margin:0cm;
font-size:11.0pt;
font-family:"Calibri",sans-serif;}
a:link, span.MsoHyperlink
{mso-style-priority:99;
color:#0563C1;
text-decoration:underline;}
.MsoChpDefault
{mso-style-type:export-only;}
@page WordSection1
{size:612.0pt 792.0pt;
margin:70.85pt 3.0cm 70.85pt 3.0cm;}
div.WordSection1
{page:WordSection1;}
--></style></head><body lang=ES link="#0563C1" vlink="#954F72" style='word-wrap:break-word'><div class=WordSection1><p class=MsoNormal><o:p> </o:p></p><p class=MsoNormal>Hello,</p><p class=MsoNormal><o:p> </o:p></p><p class=MsoNormal>First of all thanks for DigiKam, an amazing software that I was trying to find for years.</p><p class=MsoNormal><o:p> </o:p></p><p class=MsoNormal>I detected that Digikam is running all the face recognition operations on the CPU (<a href="https://bugs.kde.org/show_bug.cgi?id=431115">https://bugs.kde.org/show_bug.cgi?id=431115</a>). I’d like to improve it a bit adding GPU support. I’m working as a C# developer, but I studied my Computer Engineer degree using C, so I think I’m able to undertand the current implementations using C++. I also have a bit of knowledge about neural networks, image recognition and these kind of things because I took the Andrew Ng course.</p><p class=MsoNormal><o:p> </o:p></p><p class=MsoNormal>The first thing that I have in my mind is adding a new option like the YoloV3 one, but with GPU support (something like YoloV3 GPU – experimental). Then change the OpenCV build file to add CUDA support and then change this line <a href="https://invent.kde.org/graphics/digikam/-/blob/master/core/libs/facesengine/detection/opencv-dnn/dnnfacedetectoryolo.cpp#L89">https://invent.kde.org/graphics/digikam/-/blob/master/core/libs/facesengine/detection/opencv-dnn/dnnfacedetectoryolo.cpp#L89</a> to use CUDA instead of using the CPU.</p><p class=MsoNormal><o:p> </o:p></p><p class=MsoNormal>My main concern is this answer to that ticket from Gilles Caulier</p><p class=MsoNormal><o:p> </o:p></p><p class=MsoNormal>“Yes, opencv is not compiled with CUDA support.</p><p class=MsoNormal><o:p> </o:p></p><p class=MsoNormal>In fact plenty of features are disabled :</p><p class=MsoNormal><o:p> </o:p></p><p class=MsoNormal><o:p> </o:p></p><p class=MsoNormal>Why ? because Opencv is a biggggg puzzle, very complex to stabilize and optimize (from the digiKam side implementations).”</p><p class=MsoNormal><o:p> </o:p></p><p class=MsoNormal>So I think this change isn’t easy, due to there is no way to have both openCV instances and if I add CUDA support I’ll break other things.</p><p class=MsoNormal><o:p> </o:p></p><p class=MsoNormal>My current setup Windows/Linux with a NVIDIA 1060 GTX give me the chance to test the change in both Linux and Windows.</p><p class=MsoNormal><o:p> </o:p></p><p class=MsoNormal>Thank you for your time,</p><p class=MsoNormal><o:p> </o:p></p><p class=MsoNormal>Jorge Duran.</p></div></body></html>