PhotoShelter Unveils AI Visual Search Capability Eliminating Need for Metadata

Digital asset management platform PhotoShelter has unveiled a new feature called AI Visual Search, giving brand subscribers the ability to search their entire brand library based on visual descriptions and eliminating the need to manually add metadata to images, according to a news release.

Manually searching DAM platforms for the right visual asset can take hours, particularly for larger organizations whose brand library may contain hundreds of thousands of visual assets. PhotoShelter said 40% of the content on its servers had no metadata.

“Even the most advanced content organizations struggle to set up and maintain a comprehensive metadata strategy,” PhotoShelter said. “It would take an estimated 200 days to manually tag every image uploaded to PhotoShelter each month, and metadata-less search eliminates this incredible time-consuming challenge.”

AI Visual Search uses language and image recognition to “intelligently match” search terms to visual qualities, according to the announcement. 

“What was once a manual, difficult and time-consuming task is now a prime use case for leveraging AI to maximize efficiencies and bring better content to market,” said CEO Andrew Fingerman. “And the results are breathtaking – it’s like every conceivable keyword and caption has been applied automatically to our customers’ images.”

PhotoShelter noted that it recently launched a formal Ethics Board to routinely consider the ethics of its AI usage, improve training models, and remove any unconscious bias. “The company will also be aligning closely with the Content Authenticity Initiative to track source images and determine if AI has been used in their creation to monitor and get ahead of disinformation and authorship issues,” PhotoShelter said.

AI Visual Search joins several other AI-based PhotoShelter solutions addressing the challenges of tagging visual assets with metadata: 

ObjectID automatically creates metadata tags for images based on the objects in them, for example “football” or “X-ray.” BrandID, PeopleID, and RosterID tag people, player rosters, and brands such as a NFL team, a high school baseball player, or a coach or other face familiar among the subscriber’s asset library.

Learn more at Brands.PhotoShelter.com.


About the Author

Kristal Kuykendall is editor, 1105 Media Education Group. She can be reached at [email protected].


Featured

  • a glowing gaming controller, a digital tree structure, and an open book

    Report: Use of Game Engines Expands Beyond Gaming

    Game development technology is increasingly being utilized beyond its traditional gaming roots, according to the recently released annual "State of Game Development" report from development and DevOps solutions provider Perforce Software.

  • abstract representation of equity at the core of AI

    Why Equity Must Be a Core Part of the Conversation About AI

    AI is an immensely powerful tool that can provide customized support for students with diverse learning needs, tailoring educational experiences to meet student’s individual needs more effectively. However, significant disparities in AI access and digital literacy skills prevent many of these same students from fully leveraging its benefits.

  • Man wearing headset working on a computer

    Internet2: Network Routing Security and RPKI Adoption in Research and Education

    We ask James Deaton, vice president of network services, about Internet2's initiatives and leadership efforts to promote routing security and RPKI adoption in research and higher education networks.

  • network of transparent cloud icons, each containing a security symbol like a lock or shield

    Okta, OpenID Foundation Propose New Identity Security Standard

    Okta and the OpenID Foundation have announced the formation of the IPSIE Working Group — with the acronym standing for Interoperability Profiling for Secure Identity in the Enterprise — dedicated to a new identity security standard for Software-as-a-Service (SaaS) applications.