01 / 16
Patterns

Every wild animal
wears a fingerprint.

We read it at scale.

An engine that recognises an individual wild animal from a single photograph, turning every visitor with a phone into a sensor for conservation.

Conservation data, from one photo Live on iOS & Android 2026
The problem
2 in 5of Africa's giraffe, gone in a single generation, a silent extinction almost no one was counting.

We are losing wildlife faster than we can count it.

  • Giraffe: ~140,000 left, down roughly 40% in 30 years, with 90% of habitat lost. Some subspecies are now critically endangered.
  • Conservation still counts individual animals largely by hand: slow, costly and fragmented. You cannot protect what you cannot count.
  • The world is finally demanding credible biodiversity data: TNFD, 1,200+ nature-positive commitments, and almost no one can supply it at the level of the individual animal.
The populations

Four species, heading in different directions.

About 140,000 giraffe remain, up from under 100,000 in 2016, first because the data improved, and now because conservation is genuinely working: three of the four species are increasing. The longer story is the range itself, nearly 90% of it already gone.

SpeciesStronghold parksWild estimateTrend
SouthernKruger ~12,400 · Etosha / Namibia~69,000▲ Rising · +50% in 5 yrs
ReticulatedKenya · Laikipia, Samburu~20,900▲ Increasing · recovering after a long decline
MasaiSerengeti & Tarangire, Tanzania~43,900► Stable · holding after steep historic loss
NorthernNiger · Chad · Uganda~7,000▲ Increasing ~20% · Kordofan, Nubian & West African subspecies still critically small

Where protection is funded, tracked and acted on, giraffe recover, that's the lesson of the last decade. The ones still slipping away are the ones we cannot yet see clearly. The difference is data, and that's exactly what Patterns produces.

Kruger National Park alone holds ~12,412 giraffe (range 10,345 to 14,554), the largest single protected population. Sources: GCF State of Giraffe 2025; African Journal of Wildlife Research 2025.

Their range
Map of Africa: giraffe historic range in gold and remaining current populations in grey

Historic range vs current populations · Source: IUCN / GCF

A range became fragments.

The gold range is where giraffe once had room to move. The grey is where populations remain today, smaller and separated, hard to read from occasional surveys alone.

~90%
of historic giraffe range, lost.
What remains has to be tracked population by population.
Historic range: the savanna belt they once filled Current populations: isolated surviving strongholds
This is why live data matters. Verified sightings, individual IDs and movement records show which isolated groups are growing, where corridors still work, and where protection can restore range.
Why now

A coat is a fingerprint. A phone is a sensor. Carry the wild home with you.

  • Every giraffe, zebra, whale and leopard carries a unique, lifelong marking pattern.
  • Modern computer vision identifies an individual from one photo, in under a second.
  • Millions of wildlife photos are taken every year. Nobody is structuring them.
  • Citizen science already rewrote the map. eBird, iNaturalist and Happywhale turned everyday photos into the largest wildlife datasets on Earth. Giraffe numbers were revised upward only because the data improved.
The engine

One photo in.
The exact individual out.

Every photo runs a multi-stage computer-vision pipeline: instance segmentation isolates the animal, the coat is lifted into hundreds of scale-invariant keypoints and encoded as high-dimensional descriptors, matched against tens of thousands of references with approximate-nearest-neighbour vector search, then confirmed by geometric, ratio-tested verification and a calibrated confidence score.

01 ISOLATE
Segment
Instance-mask the giraffe from the bush.
02 ENCODE
Read the coat
Hundreds of scale-invariant keypoints.
03 SEARCH
Vector match
Across tens of thousands of references.
04 VERIFY
Confirm
Geometric ratio test + calibrated score.
Sub-second inferenceSelf-learning catalogueConservation-grade verificationLive on iOS + Android

Detection and matching run as cloud inference workers against a warm, in-memory vector index; the catalogue is the source of truth and sharpens with every verified sighting. The methodology builds on the field's gold-standard work, Microsoft's AI for Earth wildlife re-identification and the Wildbook research network, advanced into a real-time engine and tuned on live African field photography, built to a standard that makes rebuilding it pointless.

PATTERN MATCH ENGINE98.2% confidence
Coat-pattern keypoints matched between two giraffe sightings
Your photoKnown · Thabo #07
match confidence

          

          
A real coat-match from the engine. Every line is one keypoint matched to a known individual.

        

      

    

  


  
  

    

      
The hard part is done

      
The hard part is built. The next hard part is reach.

      
What looks like a photo app is a multi-stage computer-vision pipeline, a continuously-learning catalogue, a data-integrity backbone and an engagement system, each a serious project in its own right. That is the hard part, and it is already built and live.

      

        
    
          
  • The dataset compounds. Every sighting curates a living record of who, where and when. It grows richer and more valuable with every photo taken.
    • 
          
  • Field-grade accuracy, earned. Identifying an individual from a real safari photo, half-behind a tree, is genuinely hard. Ours is tuned and calibrated against a real catalogue.
    • 
        

        
    
          
  • Conservation-grade integrity. Every record is human-verifiable and auditable, not a fun guess. That's what parks and researchers can actually trust.
    • 
          
  • The engagement engine. Citizen science fails on participation, not technology. The collectible, the reveal, the joy. That's what gets people contributing.
    • 
        

      

      
Building the engine took years. The next mountain is reach, getting Patterns into the hands of enough citizen scientists to feed it. That's exactly where the right partners come in.

    

  


  
  

    

      
The reframe

      
Not a safari app. A wildlife-data platform with a consumer front-end.

      

      

        
The app is the cheapest data-collection sensor network and the most shareable conservation story in the category. The value is the structured, individually-identified sightings it produces, the data parks and researchers have never had.

        
Every serious wildlife-ID effort, from Happywhale to Wildbook, exists to serve conservation, not to sell downloads. Patterns is the same: the app is the means, the species is the point.

      

    

  


  
  

    

      

        

          
The relationship

          
Your giraffe, for life.

          
Most safari apps end at the gate. Patterns is just getting started. The giraffe you meet today becomes yours to follow, long after you've driven home.

          
One sighting becomes a lifelong thread: the giraffe, the guest and the park, all on the same map.

        

        

          
MEETPhotograph a giraffe and meet the individual: named, carded and added to your collection.

          
FOLLOWWe ping you when your giraffe is seen again, anywhere on the network, by anyone.

          
WATCHSee where it roams across months and seasons: a living movement story, not a one-off snapshot.

          
RETURNEvery ping is a reason to come back, and a reason to care about the species, not just the safari.

        

      

    

  


  
  

    

      
Where the data goes

      
The data belongs to the giraffe. So it is free to those who protect them.

      
Every verified sighting, location and movement track is handed to giraffe conservation, parks and reserves, free of charge. No paywall on the science. That's the whole point.

      

        

          
A self-sustaining loop

          
    
            
  • Guests engage for the joy, the cards, the chase, their own giraffes, and generate data as they play.
    • 
            
  • Operators and reserves carry the platform as part of the guest experience they already sell.
    • 
            
  • Conservation gets the data free, and better data means better protection, and a better story guests come back for.
    • 
          

        

        

          
It plugs into what you already run

          
    
            
  • Singita and African Parks run EarthRanger; SANParks and Peace Parks run Postcode Meerkat; &Beyond runs collar telemetry.
    • 
            
  • Patterns adds the layer they lack: individual-ID and citizen collection, and white-labels for any operator or reserve.
    • 
          

        

      

    

  


  
  

    

      
Why us

      
The science is proven. We built the solution.

      

      

        

          
The science is settled

          
Researchers already prove a giraffe can be identified by its coat: GiraffeSpotter and Wildbook, backed by the Giraffe Conservation Foundation. That part works.

        

        

          
We built what was missing

          
A consumer-grade app that turns anyone with a phone into a giraffe spotter, with the joy that keeps them coming back and the data pipeline conservation actually needs. It's built, and it's live.

        

      

    

  


  
  

    

      
Built and live

      
It's built. It's live. It works.

      
Detection, coat-mapping, the live match index and the signed-photo pipeline are built, deployed and running today, identifying real giraffe from a single photo.

      

        
    
          
  • Live app on iOS TestFlight and Android.
    • 
          
  • Production backend, real coat matches at high confidence.
    • 
          
  • Brand and marketing site live (patternsproject.co.za).
    • 
        

        
    
          
  • What remains isn't engineering. It's partnerships: with corporates, conservation initiatives, and private and public game reserves.
    • 
          
  • The app can be white-labelled for an operator, a reserve or an NGO.
    • 
          
  • That work is relationships and trust, not code.
    • 
        

      

    

  


  
  

    

      
Where we are

      
The technology is built. The product is ready. Now we get it into the right hands.

      

      

        
Built
The engine, the app and the data pipeline are live on iOS and Android. The hard part is done.

        
Ready
A polished, conservation-grade product, ready for real giraffe, real parks, real guests.

        
Now
Get it into the hands of the people who will love using it, and whose sightings power the data.

      

    

  


  
  

    

      

        

          
Choosing the right partners

          
Conservation first. Everything else follows.

        

        

          
Patterns is, first and foremost, about protecting wildlife and getting the next generation to fall in love with the bush and the creatures in it.

          
Commercial branding has its place. But the goals that come first are driving usage and knowledge, and protecting the beauty of nature, and those are the goals we'll share with any partner who joins us.

          
We built the solution. Help us get it into the field, and let's save giraffes together.

        

      

    

  


  
  

    

      
Hold in reserve

      
Appendix

      
Numbers and sources, and the caveats.

    

  


  
  

    

      
Appendix · numbers & sources

      
Every figure, with its source.

      

        
Giraffe ~140,000, down ~40% in 30 yrs, 90% habitat lost · Giraffe Conservation Foundation.

        
By species: Southern ~69k, Masai ~44k, Reticulated ~21k, Northern ~5.9k · GCF State of Giraffe 2025.

        
Southern giraffe up ~50% in five years · GCF State of Giraffe 2025.

        
Kruger ~12,412 giraffe (10,345 to 14,554) · African Journal of Wildlife Research 2025.

        
West African giraffe 49 (1996) to ~600 today · GCF / Operation Sahel.

        
Numbers revised up from <100,000 (2016) as data improved · GCF.

        
Wild Me / Wildbook: nonprofit, 8 staff, 53 species · Wild Me, National Wildlife Federation.

        
GiraffeSpotter / Wildbook for Giraffe · Giraffe Conservation Foundation, giraffespotter.org.

        
Computer-vision approach · Microsoft AI for Earth, Wildbook / Wild Me.

        
Citizen science at scale · eBird (Cornell Lab), iNaturalist, Happywhale.

      

      
Note: population figures are best-available estimates from GCF and peer-reviewed assessments; ranges are given where published. Full links in the companion memo, business-case-response.md.

    

  












← → to navigate · F fullscreen · P to print