Development · integration · reverse engineeringOstwald · Alsace · Remote projectscontact@dtouch.fr
Elasticsearch · photo library · quality

Search, indexing & deduplication of images

A production photo library must be able to find the right resource among originals, previews, editings and archives. The journey Reymann includes indexing the IMP and searching for duplicate images.

Index content to find it

The Rey’actif ecosystem used Elasticsearch for indexing PIM text and image data. This search layer completed the content storage and had to remain consistent with the repository.

A useful search depends on the information available: references, designations, media links and metadata. The scoping consists of understanding how teams are looking and what allows them to identify the right result.

A duplicate is not always an identical file

The channel may contain a high-definition original, a preview, a cut-out version and a variation intended for the web. Storing them as duplicates without examining their role would cause a loss of production.

Diagnostics included high and low definition relationships, unoriginal previews, and aberrant images. It was necessary to distinguish a technical deviation from an expected variant.

Search similar images

A duplicate image search tool with AI was included in the production environment. It complemented file-based and metadata-based controls.

The result of a detection must be examined in its context: product reference, edition, clipping, dimensions and quality of the original. Visual similarity helps find candidates; Conservation decisions depend on the actual role of resources.

Maintaining links with documents and archives

An image that is moved or misreferenced may remain present on disk and become unobtainable for publication. Recaptures and XML exports also required checking paths and correspondence.

We link the diagnosis of the photo library to its uses: search, consultation, page preparation and export. The quality of the fund is measured by the ability to reuse the right image in the right place.

Understanding the exchanges

An image returns its product, versions and uses.

The organization of the photo library combines identifiers, metadata, preparation of visuals and reuse in publications.

An image returns its product, versions and uses.Enlarge the diagram
  1. 01

    Entry of visuals

    Images and supplier data

    Reference • radius • SKU
    Identify resources ↓
  2. 02

    Identification

    Bringing the product closer

    Nomenclature • metadata
    Preparing for Decline ▼
  3. 03

    Preparation

    Create useful resources

    Clipping • framing • versions
    Check returns ▼
  4. 04

    Control

    Check resources

    HDEF / overview • profile • duplicate
    Sorting resources ▼
  5. 05

    Photo library

    Preserve and find

    Reference • variant • condition
    Publish and reuse ▼
  6. 06

    Uses

    Reuse the right files

    PIM • pages • web supports
An image returns its product, versions and uses.Enlarge the diagram
Functional diagram · synthetic reading of flows and control points.
Closer

Distinguish between the product, the original file and its variations.

Control

Identify duplicates and gaps between previews and high definition.

Reuse

Find the appropriate visual for the document and its destination.

Your job deserves
the right tools.

A specific need, a software to evolve, a project to frame.

Exchange on your project

Complementary expertise, the same high standards.