Blog / Foundations, Metadata

Tagging People in Images

Peter Krogh
Fri Sep 18 2020

Tagging media for the people pictured is one of the most valuable bits of metadata you can add. There are several methods to use and several places that people tags can be saved.


Keywords are a great place to save your people tags. They are broadly readable, and fast to apply. If you are using a tool with hierarchical keywords, then you can organize people into groups: family, company division, assignment, department, etc. Many keywording tools will have some capability to enforce a controlled vocabulary list to ensure consistent naming.

IPTC Person Shown

The person shown field in the IPTC Extension is a field that is specifically devoted to naming people in a photo. This removes ambiguity that can sometimes arise when a picture is about a person, but not of a person. It also introduces some ambiguity, however, because it does not support a hierarchy to help you understand which John Smith this is.


The caption or headline field can also note the name of a person shown in an image. For images that are being sent out for publication, it’s customary to identify people in the caption field. When more than one person appears in a photo, the caption should allow you to know which person is which. A caption gives you the ability to remove any ambiguity as to which John Smith is in the image, since you can add a description of the person.

Machine Learning

Computer assisted face tagging is one of the earliest uses of ML technology for images. We see it in many applications. In all cases, the system needs to be trained to recognize particular individuals. While not perfect, it’s becoming very accurate.

Region Tagging

“Region” tagging metadata supports drawing a box around a face and attaching a person's name to the face. Depending on how the software is designed, the face tagging tool can also write the name to the Person Pictured field and/or the keywords field. Many ML tagging services will also make use of regions.

Black Box

Black box DAM typically offers face tagging, although the tags may not be written back to the files. And as with any ML tagging, it must first be trained to recognize individuals.

Custom Metadata

Custom metadata fields may be useful for some person tagging as a tool within an application. The IPTC Person Shown field does not support hierarchy, so it may become difficult to use with large lists of people. Creating a custom metadata field that allows you to sort people into categories may allow for more clarity and efficiency compared to a giant flat list. However, when a custom metadata field is used, it would still be desirable to write the person tag to one of the more widely-supported fields like keywords or person shown upon export. This prevents the tag from becoming orphaned in an unsupported field.

Name Formats

You should decide on a way to write names and stick to it. It could be Firstname, Lastname, or the other way around. If you are using keywords, you could also use a hierarchy to better define who individuals are (especially those with the same names).


If your archive contains multiple people with the same first and last names, you will probably want to use some method to distinguish between individuals. If there is a middle name or initial, that can be useful to distinguish individuals. Likewise if there is a prefix or suffix (e.g. Dr. or Jr.) that can also be used to provide uniqueness. Archivists will make use of birth/death dates to distinguish individuals, in the form FirstName Lastname (Birthdate-Deathdate). You could also use some other distinguishing role or profession to define a particular person, such as Paul Simon (musician) and Paul Simon (Former Senator). Note that parenthesis are typically illegal characters in keywords and using them may cause unpredictable results as files move from one service to another.

Is Mediagraph right for your organization?

Let’s find out together.

Book Your demo today