I'm using Adobe Elements Organizer 2019 since e few months. I imported my 97000 photo's and video's and turned on Face Recognition. This took a while but my iMac did it and now all faces are somehow found by the program. However my problem is that the first stacks were quite large with sometimes more than 100 people in it so this was working fine, of course still al lot of work because the software was not always right in choosing the right combinations. But after a lot of evenings and Sunday afternoons I got to the point when I thought this is the last stack and then there was the small selection box in the top left corner saying 'Hide Small Stacks'. When I unchecked this box there popped op over 20000 stacks of peoples faces. The stacks contain only 1,2 or sometimes 3 faces so it is almost undoable to manually organise these enormous amount of faces.
When I look at the stack I wonder why the software did make so many stacks because the faces of a lot of people ar not so different that you would choose to put them in separate stack.
Now is my question is there any way te restart the analyses of the stacks which the peoples faces that are not yet assigned to an name? Of course the already assigned people and stacks should not be reanalysed. I hope there is a way otherwise it will take me a year to organise the faces and photos.
Paul, I'm afraid that if you are as obsessive-compulsive as I am about face recognition, it is going to take you a while to name all your stacks. While it is possible to reset the analysis in Preferences, this will remove all existing face recognition and start the process again.
What you are seeing is typical in my experience. I am currently working on a few thousand small stacks. Many of the stacks are of my closest family members for whom I have more than 10,000 named faces each. Yet, the auto-analysis fails to recognize them and, as you have seen, fails to place numerous images in one stack. I sometimes see a raw+jpg shot placed in different stacks.
Now, having said that, there are some strategies to speed up the process. First, I select a group of stacks like this screenshot with a Shift+Click (or Ctrl+Click for non-contiguous stacks). This brings up a Merge People button at the bottom instead of the Add Name button. The Merge People choice is also available via context menu. Once the button is selected, a text box comes up for naming the selected person.
You will also probably find that many of the single stacks are either not faces or are not people you want to name. With these, I try to name the people I want to name who are interspersed in the stacks. Then, I select 30 or 40 stacks at a time and choose the Don't Show Again button.
Face recognition isn't perfect and you will find probably find many faces that have not been recognized at all. So, try not to go too crazy over this process.
Hi, a bit old thread but apparently still an issue. To me it seems like it recognizes the faces well enough but the combining of stacks fails at some point and never attempts to compare faces that at least to me appear like almost identical. This produces a huge amount of 1 pic stacks that have to be gone through manually. Would love to see some improvement in this even without upgrading to 2020 version since I don't see that much benefit in upgrading.
@samih70364729, in my experience (with a 200K media catalog), Elements 2021 face recognition performance has greatly improved over prior version, both in accuracy and speed. In particular, much larger stacks are generated with the latest version and correct names are auto-suggested for most stacks of new photos of my family members. False positives are also much fewer. Also, the latest version face recognizes video files. If face recognition is important to you, I would recommend the upgrade. You can try it out for yourself with a 30 day free trial download.
Thanks for you reply, I was expecting something like this. So we have to be patient and work our way through the pile of photos. Thanks for the tips, I found it out myself already to work in the same way as you described.