Hi everyone,
I’ve been dedicating significant time to exploring the “People” section of digiKam, and I’m seeking best practices on a few topics to make informed decisions with the current state of face recognition. I believe this discussion could greatly enhance the manual, and I’d love to contribute to it.
Infancy / Toddler / Childhood / Adolescence / Adulthood
Is it necessary to create separate tags for each developmental stage?
How does the algorithm handle it if all growth stages are tagged as one?
I know that the following three questions are partially covered in the manual, but I would appreciate further clarification.
When not to tag faces?
While tagging, is it advisable to avoid tagging faces under these conditions? Please share your insights on each point:
- Wearing sunglasses
- Having a beard
- Significant weight gain
- Partially shown (e.g., only one side visible, obstructed by an object)
- Making funny faces
It’s clear that carnival makeup doesn’t help with face recognition.
Blurry or not sharp faces
Does tagging blurry or unsharp faces assist or hinder the algorithm in recognizing such faces in the future?
Number of Faces for Comparison
Manual states atleast 5 but how many face samples should be tagged to improve the accuracy of face recognition? Is there an optimal number of faces per individual to ensure reliable recognition?
Pet Face Detection
Does digiKam support face recognition for pets? If so, what are the best practices for tagging and managing pet faces? How does the algorithm handle different breeds and species?
Correction of Incorrect Tags
What is the best way to handle and correct incorrect tags? How does the algorithm learn from these corrections?
Integration with Other Software
How well does digiKam’s face recognition integrate with other software?
