Visual entity resolution · Patent pending · Your environment
Find every photo of a person. Across a lifetime.
PhotoGraph turns an unlabeled photo collection into a searchable identity graph. Start from one picture and get back every appearance of that person: childhood slides, crowded group shots, faces in shadow, even photos taken from behind.
And once you have that graph, it’s the foundation for so much more: layering on context terms, semantic analysis, property details from your metadata, and revealing the degrees of separation between two or more people (the Kevin Bacon game). All of it starts with knowing you have entities and their context. We give you that.
- No labeling or model training
- Runs on your hardware
- Results as a queryable graph
- 524 of 524In a rigorous benchmark study, PhotoGraph found 524 images of a person, versus a widely deployed model returning 81 at most. The next best performer found only 240 at most.
- Age 2 → 76Real-world example: one person grouped across 74 years of film, slides, prints and digital.
- FastFind every photo of one person in thousands or millions. Time-to-market: no training needed, so you get value fast.
- 0 lossImages and results never leave your environment. You have total control.
Beyond the face
We don’t do facial recognition. We make it useful.
Facial recognition asks: Is this face a match?
PhotoGraph asks: Is this person anywhere in this collection?
- Entity resolution, not face matching
Facial Recognition compares one image to the collection. Entity Resolution compares every image to every other image.
- A whole lifetime
Groups a person from a faded childhood slide to a grandparent’s snapshot, and doesn’t confuse look-alike relatives.
- Photos from the real world
Finds people who are behind objects, low resolution, poorly lit, poorly focused, and even turned away from the camera.
We find the photos other platforms miss.
Real results from one family collection: the same person, surfaced in the shots other tools skip.
Other photo apps find
Front-facing, well lit, posed.That’s where facial recognition stops.
PhotoGraph also finds
Age 3, faded slide 
Face hidden by a hat 
Backlit, in shadow 
Photo of a photo 
Contact-sheet frame 
Facing away 
Hand over face 
Behind a toy tower
How it works
From an unlabeled archive to an identity graph in four steps.
Nothing to label, nothing to train. PhotoGraph works on your collection from the first day of deployment.
- 01
Connect to your collection
Point PhotoGraph at scanned prints, slides, negatives, contact sheets, and digital photos on your own storage.
- 02
Extract individuals in the photographs
PhotoGraph finds every person in every photo, including faces in group shots, people partly hidden and people turned away from the camera.
- 03
Resolve the individuals
The Graph Resolution Core links every appearance of each person into one identity, across age, image quality and pose.
- 04
Search by example, and put it to work!
Start from one photo to see every appearance of that person. Then query the identity graph through the API, feed your catalog or search product, or ask questions in plain English.
Use cases
Built for collections where who matters.
Give subscribers every photo of their ancestor.
Genealogy platforms hold billions of images with no person-level navigation. PhotoGraph finds every appearance of the same individual: infant to elder, across donated archives where names and captions were never attached.
- A new subscriber feature: “show me every photo of her” from a single picture.
- Stranded inventory becomes searchable and starts earning.
- Works without names or captions, so uncaptioned uploads count too.
Connect people across every donor collection you hold.
Historical societies, libraries and museums can surface and cross-link individuals across their entire holdings. Life stories that were split across boxes, donors and decades come together on their own.
- Visitor family search that starts from a single photo.
- Exhibitions built around people, not just places and dates.
- Faster cataloging: seconds per person instead of weeks of manual review.
Identity resolution that never leaves your secured environment.
Organizations with strict data-sovereignty rules can deploy PhotoGraph inside their own perimeter. Link appearances across case imagery, including partial and obscured views, with no cloud service involved.
- On-premises only: no image or result is sent outside your environment.
- Works on imperfect imagery: obscured, low-resolution or turned away.
- Results as a graph that analysts can audit and query.
Turn a legacy photo library into a person-indexed archive.
Studios, stock photo companies and media companies hold decades of photography indexed by date and assignment, not by who is in the frame. PhotoGraph adds the person index.
- Rights management: find every image of a person before you license it.
- Licensing and discovery: surface archive shots editors didn’t know existed.
- Stays on your infrastructure, alongside your existing asset system.
- Evelyn HartIn the caption
- Marcus ValeFound by PhotoGraph
- Lena OrtizFound by PhotoGraph
- Theo WhitcombFound by PhotoGraph
The performance gap
Revenue-generating content found with us, lost by others.
Every photo a method misses is content your customers never see. In a real 33,000-photo family collection, one person appears in 524 photos. PhotoGraph found all 524. A bleeding-edge algorithm found 240. A widely deployed model found only 81.
Which method will find the most photos of this person?
The real-world test: a true 33,000-photo family collection. One known person appears in 524 of them, ages 2 to 76, film scans to digital, solo and in groups.
For perspective: with other methods, you see only a fraction of this person’s photos, and miss what could be the heart of the narrative somewhere among the remaining ~33,000 images. With PhotoGraph, the job is done.
Deployment
Your photos stay yours. By design, not policy.
PhotoGraph was built for organizations that cannot send images to a cloud service, whether because of law, regulation, contracts with donors, or simply because the collection is the business.
On-premises
Runs entirely inside your environment, on hardware you already own. No images and no results leave it.
Queryable identity graph
Every result is a structured graph your teams, tools and AI systems can query and build on.
Lean on compute
Far less compute and energy than other approaches, so your large collections can resolve in hours rather than days.
Questions
Get to know PhotoGraph
What is visual entity resolution?
Entity resolution decides which records refer to the same real-world thing. Visual entity resolution does that with images: it decides which photos show the same person and links them into one identity, even when nothing in the metadata connects them.
How is this different from facial recognition?
Facial recognition compares two faces and returns a match score. It needs a clear, front-facing face. PhotoGraph resolves a person across a whole collection, including photos where the face is partly hidden, shadowed or not visible at all, and across decades of aging.
Do we need to label photos or train a model?
No. PhotoGraph works from pixels alone. There is no labeling, no training set to build and no model training on your collection.
Where do our photos go?
Nowhere. PhotoGraph is deployed inside your environment and runs on your hardware. Images and results stay there.
How do we get started?
Most customers start with a pilot on part of their own collection, so they can judge results on photos they know. Ask about a pilot.
Get started
See PhotoGraph on your own collection.
Bring a set of photos you know well. We’ll show you who else is in them.