<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body dir="auto">
<div dir="auto">Hi Steve, </div>
<div dir="auto">For face detection (is this a face?), the model is fixed. No additional training is done by the end user's actions.</div>
<div dir="auto"><br>
</div>
<div dir="auto">There are even more improvements coming in 8.6.0 for both face detection and recognition.</div>
<div dir="auto"><br>
</div>
<div dir="auto">Cheers,</div>
<div dir="auto">Mike</div>
<div dir="auto"><br>
</div>
<div dir="auto"><br>
</div>
<div dir="auto"><br>
</div>
<div id="composer_signature" dir="auto">
<div style="font-size:12px;color:#575757" dir="auto">Sent from my T-Mobile 5G Device</div>
</div>
<div dir="auto"><br>
</div>
<div><br>
</div>
<div align="left" dir="auto" style="font-size:100%;color:#000000">
<div>-------- Original message --------</div>
<div>From: Steven Robbins <steve@sumost.ca> </div>
<div>Date: 1/12/25 4:10 PM (GMT-05:00) </div>
<div>To: digikam-users@kde.org </div>
<div>Subject: Face Detection in DK 8.5 - does it learn? </div>
<div><br>
</div>
</div>
First, I will say again that I am over the moon with the improvements in <br>
version 8.5. Face detection & recognition is orders of magnitude better than <br>
in the past.<br>
<br>
I just have a small question to clarify my understanding on the detection part <br>
-- i.e. when the face rectangles are created.<br>
<br>
I am using the YuNet model, with default parameters (Accuracy = 7, Face Size = <br>
Large). I find a fair number of "non face" regions identified. The manual <br>
advises to use the "Remove this face region" option. Which is fine: the <br>
rectangle is removed and I'll never be bothered again for this picture (unless <br>
I use options to re-scan the picture).<br>
<br>
My query is whether removing a face region will affect future model accuracy in <br>
any way. For example, there are often many similar pictures of the same <br>
scene. If a particular tree trunk is detected as a face, it shows up in all <br>
pictures. If I were to remove the region, is there any update to the model <br>
that reduces the chance that the particular tree trunk will be detected as a <br>
face in a future image of the same scene?<br>
<br>
Thanks,<br>
-Steve<br>
</body>
</html>