Photo tagging
Puts reviewed keywords on a shoot as it lands, so a decade-old library is findable by the words a client actually says out loud.
Tools
Amazon Rekognition, Google Cloud Vision, Adobe Lightroom, Photo Structure
Outcomes
Keywords reviewed before they reach the library • Wrong labels caught inside one folder • Search answers to the words clients use • Each new shoot labeled on arrival
Documentation
Instruction-ready detail below
Machine labelling is fast and confidently wrong. It calls confetti a birthday, tags eight hundred near-identical dance floor frames, and writes a landmark nobody will ever search for. A photographer who trusts that output ends up with a library that cannot be searched at all, because the words that matter were never in it. This workflow takes the suggested labels for a new import, pulls them into a list against the file name, drops the ones that miss, and offers replacements drawn from the terms the photographer already uses. A sample of frames with the chosen labels goes back for a person to confirm before the batch is written to the library, so a bad run costs one folder instead of a season. The photographer still decides which words matter, what a frame from this job is called, and when the older years get their turn.