Immich is a great tool that i can recommend, they have ML tools to analyse photos and make them serachable by image content

I store all my photos, along with other digital things, on a Synology NAS that has a decent built in photos organizer that lets you do some level of searching, sorting by people, tags, etc.

A likely better, but not free-with-NAS-hardware, option would be something like Excire Foto or one of its many competitors.

I happened to choose ACDSee Photo Studio to add keywords, ratings and catalogue them, so I could do the simple “Show me all pictures of Magpies with a Rating of 4” searches (because I want to find my best Magpie photo, etc.). Product chosen as it was not subscription, so works “forever” (though you do get bombarded with upgrade offers).

The keywords show up as tags in Windows File Explorer properties, and you can now add tags there directly. Don’t go for any approach that is not visible in your operating system, outside of the app!

I’m not sure it was the best option, but, as purchased, it does the job I needed. Ironically I ended up with the full ACDSee package, for all I only needed the cataloging form, and in fact do my small amount of post-processing just in Canon DPP4.

Relying on a file system to hold meta-data about images is bad practice. Never use file names, folder structure, or folder names to be the sole carrier of information. Image files themselves hold meta data, and it is best to make use of it. That way, your image files are truly portable, and contain everything that might need to be known if they become divorced from the file system.

In the past, a convention was to tag images with keywords. This was a tedious process and many photographers spent months and years carefully keywording and hierarchically organising keywords in order to feel that they could retrieve whatever they wanted with a simple search.

In reality, for all those years spent keywording, few photographers leveraged the keywords. In other words, they maybe spent a year doing complex keywording and never recouped that time with future searches, because after all that work, they only ever did a few searches. (“every image with me and a flower in it”). Fortunately with AI now, it doesnt matter, there is little need to spend time keywording. And as AI automatic tagging gets better, keywords will be needed less.

For “snapshots” adding brief descriptions to the meta data is good practice, especially if the images are historic, personal, genealogical, or informational. Descriptions should follow a consistent brief template, such as one phrase for each of : what-who, where, when, physical context. For example:

Martha [Hayes] Brown and Alice [Tate] Cook celebrating their birthdays on a boat. East River near Coentes Slip. Taken by Ron Brown June 3 2014..

You might argue that the date is not necessary as it should be in the exif date tag, but it’s good to have it in the description/caption as well; some web applications only show the caption. For scanned photographs, add physical context: ie:

Martha [Hayes] Brown and Alice [Tate] Cook celebrating their birthdays on a boat. East River near Coentes Slip. Taken by Ron Brown June 3 1927. Scanned 8x12 print stamped with Hansen Photography watermark. Hand annotated on reverse: “Having Fun!” in Ron Brown’s hand.

It is important not to over-tag: for a simple image of a flower, you might simply want the name of the flower unless there is some reason viewer need to know where that flower is. Other exif data will have that information.