Asset indexation
Images get uploaded and organised the same way, every time.
AI suggests the description, and duplicates are caught before anything is saved.
Images are selected, and AI suggests a description of each one. A person reads and corrects the suggestion, and the web app, a program in the browser, checks each image automatically for duplicates, without AI. For a possible duplicate, a person chooses whether the image is skipped and not saved, or kept. Every image that is not skipped is saved to a shared image library.
Shown with example data from an invented ceramics studio.
- In and out
- Selected images in. Images with a description in a shared image library out.
- People
- Select the images, read and correct each AI suggestion, and choose whether a possible duplicate is skipped or kept.
- AI
- Looks at each image and suggests a description.
- Systems
- The web app is a program in the browser. It finds possible duplicates automatically using a digital fingerprint: a code calculated from each image. This step does not use AI. The library saves the images with their descriptions.
Without fixed steps, images are described inconsistently or not at all.
- Duplicates pile up, and a particular image is hard to find.
- Afterwards, the library has to be tidied up by hand.
People, AI and systems each have their own part.
The drawing follows one image from the moment it is selected until it is saved with its description. Each step says whether a person, AI or a system does the work.
- StartImages selectedThe person uploading picks the images.
- 1 · AIAI suggests a descriptionEach image gets a suggestion, kept as a draft.
- 2 · PeopleA person corrects itThe draft is read and corrected by hand.
- 3 · SystemsDuplicates found automaticallyThe web app compares the image's fingerprint with the library. No AI.
- 4 · PeopleSkip or keepOnly for a possible duplicate: a person chooses whether the image is kept and saved.Skip: the image is not saved.
- ResultSaved to the libraryEvery image that is not skipped is saved with its description.
What each step looks like.
Images selected
The person uploading selects the images. From there, every image follows the same steps.
For uploadExample dataImages4 selectedIMG_4127.JPGSelectedIMG_4131.JPGSelectedIMG_4135.JPGSelectedIMG_4140.JPGSelectedAI suggests a description
AI looks at each image and suggests a description of it. The suggestion stays a draft until a person has read and corrected it.
AI suggestionExample dataIMG_4127.JPGDraftLow bowl with a pale blue glaze on a white table.
A person corrects it
A person reads each suggestion and corrects it by hand. The corrected text is saved with the image.
DescriptionExample dataIMG_4127.JPGCorrectedLow bowl with a pale blue glaze
Read and corrected by a personon a white tablefrom the spring series.Duplicates found automatically
Before saving an image, the web app automatically compares it with the images in the library. It uses a digital fingerprint: a code calculated from each image. This step does not use AI. In the example, the vase is marked as a possible duplicate because the same file name and a similar image already exist.
Duplicate checkExample data- IMG_4127.JPGLow bowl with a pale blue glaze from the spring series.New
- IMG_4131.JPGVase in raw clay with a narrow neck.Same file name and a similar image already existPossible duplicate
- IMG_4135.JPGMug with a handle and a white glaze.New
- IMG_4140.JPGSmall jar in raw clay with a short neck.New
Skip or keep
If the image is a possible duplicate, a person chooses whether it is skipped or kept. A skipped image is not saved. A kept image is saved with its description.
Skip or keepExample dataIMG_4131.JPGVase in raw clay with a narrow neck.Same file name and a similar image already existPossible duplicateSkip: the image is not saved.Keep: the image is saved with its description.
Every image that is not skipped is saved with its description.
Images that are not skipped end up in a shared image library, each with a description a person has read and corrected. A possible duplicate that is kept ends up there too. That way a particular image can be found again.
- Description
- Low bowl with a pale blue glaze from the spring series.
- File
- IMG_4127.JPG
What changes in the work
Before
- DescriptionsInconsistent or missing.
- DuplicatesPile up.
- A particular imageHard to find.
- AfterwardsThe library has to be tidied up by hand.
After
- DescriptionsAI suggests one, and a person reads and corrects it.
- DuplicatesThe web app flags possible duplicates before the image is saved.
- A particular imageSaved with its description, so it can be found again.
- AfterwardsA person chooses whether a possible duplicate is skipped or kept. A kept image is saved.
Same steps, other images.
Three invented examples of the same steps: images are selected, AI suggests a description, a person corrects it, and every image not skipped is saved to a shared library.
Product photos in a shop
- Images of
- New products
- AI suggests
- A short description of the product
- A person corrects
- The person who knows the products
- Saved to
- A shared library of product photos
Photos from a building site
- Images of
- The work on site, week by week
- AI suggests
- What the photo shows
- A person corrects
- The person running the project
- Saved to
- A shared library for the project
Photos from an association
- Images of
- Meetings and events
- AI suggests
- A short description of the scene
- A person corrects
- The person in charge of communications
- Saved to
- A shared library for the association
Images that need describing and keeping in one place?
I'll go through your message, and if the job is not a fit, I'll tell you clearly.