For the 7th SFG · Eglin · NW Florida

One picture. Every angle.
Real time.

This is computer vision applied to a problem operators already understand — six camera feeds fused into a single bird’s-eye view of an engagement, running on edge hardware, with vehicles and people tracked live.

We build systems like this for a living. Helmet-cam fusion for an SF team is the next version of the same problem, and we’re the right team in the right ZIP code to build it with you.

Hailo-8 + PETRv2 · 6 nuScenes cameras · open-source reference demo

Computer vision at the edge — the building blocks

Detection, pose, depth, segmentation. These are the primitives we combine to solve fusion, tracking, and situational-awareness problems on operator-worn and operator-carried hardware. Every tile below is running on a low-SWaP edge accelerator — no cloud round-trip, no latency penalty.

Object detection

Vehicle and personnel ID at the edge

Pose estimation

Body language and intent inference

Depth estimation

3D scene reconstruction without LiDAR

Instance segmentation

Tracking individuals through dynamic scenes

Open-source demos courtesy of Hailo · github.com/hailo-ai

The automotive case is the proof. Your mission is the build.

The clip above is an open-source automotive reference — six fixed cameras fused in real-time on edge hardware. The technique is mature; the platform isn’t a car.

Adapting it to helmet-cam streams for an SF element is an engineering effort, not a research bet. The path runs through camera geometry, ego-pose without GPS, and ground-plane handling for the terrain operators actually move on — all solvable, all the kind of work we do.

That’s the conversation we want to have with you.

Why this team, for this problem

01Computer vision

Computer vision is what we do

Multi-camera CV and sensor fusion. Edge AI on purpose-built accelerators — Hailo, Jetson, and the class of devices that fit on a kit or in a ruck. Object detection, pose, depth, segmentation, tracking — deployed, not theorized. A decade-plus of applied AI/ML across multiple domains between the two of us.

02In the room

Real defense experience, both sides of the fence

INDY: Navy pilot, 10+ years USAF flight test engineer. BOOT: supported AFWERX from inside the Air Force. Between us we’ve sat across from SF operators, flight crews, and program offices — we know how the conversation goes, and we know which details actually matter to the end user. Comfortable with comms integration and ATAK plug-in workflows; we’ve used the tools you use.

03Government-ready

Your contracting shop already knows what to do with us

SAM/CAGE registered. ATO process literacy. CUI and SECRET handling. AWS and GCP — including the AFWERX GCP environment under AFSOC/AFRL. Deliver on-prem, hybrid, or cloud, depending on what your security posture needs. SOFWERX engagement in process.

Bonus — we’re neighbors. Crestview, FL. Front door to Eglin’s gate is shorter than most contractors’ commute to their own office.

Pick a 30-minute slot — we’ll come to you, or call you

No capability statement, no slide deck. First call is 30 minutes — you tell us what the unit needs, what you can share, and what shape a budget might take. We tell you what we’d build and how fast.

Book a 30-minute call →

Prefer a different channel? eric@octocore.ai · +1 919 524 8595