You're inspecting a sample, not the run
Spot-checks miss whatever happens between them. Automated vision inspects 100% of units as they pass — nothing leaves the line unseen.
Automate the Inspection.Perfect the Line.
Custom machine vision that inspects every unit, catches every defect, and optimizes your throughput — models trained on your product and your line, not a generic detector that half-works on real footage.
Book a consultation
Automated Inspection
Quality Control
Axis Partner
One team builds the model and the system behind it — 100% in-line inspection, defect detection and yield analytics that run every shift, with FSMA 204 traceability along for free.
Built on
Axis Communications
FDA FSMA 204
GDST / EPCIS
ONVIF & edge AI
CamStreamer
The cost of manual inspection
Manual quality control can't watch everything.
A person can spot-check a sample, not the whole run — and between checks, defects slip through. Eyes tire, standards drift shift to shift, and by the time a flaw is caught downstream (or by a customer) it's already expensive. There's rarely an image trail to prove what shipped. Automated vision changes the economics: every unit, every shift, one standard.
Human inspection drifts
Fatigue, subjectivity and shift-to-shift variance mean the same defect gets judged differently by different people. A model applies one standard to every unit, every time.
Defects caught late cost the most
A flaw found at shipping — or by your customer — costs far more than one caught in-line. Catching it at the source is where the money is, and where cameras earn their keep.
What we do
Three outcomes we build cameras to deliver
One custom-vision capability — trained on your product — pointed at the three things that move your numbers: automation, quality, and throughput.
Automated inspection & monitoring
Put a camera where a person can't stand all day. Every unit checked, every shift, with no added headcount and an image kept for every event.
- 100% in-line inspection, not spot-checks
- Runs 24/7, unattended, no fatigue
- Instant alerts on anomalies and exceptions
- Image evidence retained for every unit
Quality control & defect detection
Catch the flaws people miss and apply one consistent standard to every unit — then grade and sort automatically instead of by hand.
- Surface, shape and color defect detection
- Size and grade classification
- Foreign-object and contamination checks
- Assembly and label verification
Throughput & yield optimization
The same camera feed becomes operational insight — where the line slows, where yield leaks, and how one shift compares to the next.
- Count, yield and throughput analytics
- Bottleneck and downtime detection
- Shift-over-shift and line-over-line comparison
- Waste and reject-rate reduction
What automated vision changes
From spot-checks to total coverage
We're a young practice, so these aren't past results — they're the standard a well-scoped inspection system is built to hold.
100%
Units inspected, not sampled
24/7
Runs every shift, unattended
1 standard
Applied to every single unit
Every reject
Logged with image evidence
One application, in depth: passive traceability
Four steps between a camera and an audit-ready file.
The same detect-track-and-record pipeline behind every custom build, shown here on our flagship use case. No new hardware for the crew to carry, no app to remember — the camera watches the work, the model recognizes the event, and the register fills itself in.
- Events today
- 1,284
- Traceability lot
- TLC-2291
- Throughput
- 30 fps
Observe
Cameras cover the points where lots change state — receiving, grading, transformation, shipping.
Recognize
Models trained on your product identify the crate, the label, the scale reading, the operator, the movement.
Record
Each detection becomes a structured event with a Traceability Lot Code, timestamp, location and quantity — validated on the way in.
Export
When a request arrives, the register produces an FDA-ready Electronic Sortable Spreadsheet in one action.
Where we are
A new practice with working software
We're early. Rather than dress that up, here's exactly what exists today — you can look at all of it before you talk to us.
Built · Traceability platform
A full traceability platform, built for a stone crab operation
A working reference implementation of passive traceability, built end to end for a stone crab aquaculture operation: operations feed, environmental monitoring with threshold alerts, harvest history, an FSMA 204 lot register with one-click FDA export, a GDST/EPCIS event stream, and a five-layer data-quality system that catches bad records before they enter the register.
A working build, not a mockup. We tailor the same software and integration — the models,
the alert thresholds, the lot schema, the exports — to your operation. The architecture
generalizes to anything lot-based: shellfish, finfish, produce, cold-chain distribution.
Running · Live camera systems
Aviation livestreams with real-time data overlays
Public-facing camera streams built on CamStreamer and Flightradar24, burning live flight data onto the video feed and running unattended for months at a time on Axis hardware.
Publicly viewable and continuously running — the plainest evidence we can offer that
we get cameras, data sources and software to cooperate reliably in the field.
Straight about it
Tangent Solutions is a young practice. We hold Axis Technology Partner status, we've built and can demonstrate a full FSMA 204 traceability platform, and we run live camera systems in production — but we have not yet delivered a paid traceability deployment, and we'd rather you heard that from us than found it out later. What that buys you: our full attention, pilot-first pricing, and a founding-customer relationship with the person who writes the code.
Quality & grading
Automated size and grade classification, shell and surface defect detection, harvest-readiness assessment on the conveyor or in the field.
Environmental & stock health
Biomass and growth tracking, feed-waste reduction, water-condition thresholds and real-time alerting on surface and underwater feeds.
Live camera streaming
Public-facing livestreams built with CamStreamer and Flightradar24 — aviation and site cameras with real-time data overlays.
Industries
Where this applies
One custom-vision capability — detection, tracking and structured event data — applied anywhere something moves through a process. Food traceability is where the regulation bites hardest, but it's one column of many.
Manufacturing
FlagshipIn-line quality control and automated optical inspection, defect and assembly verification, throughput analytics and production-line automation.
Aquaculture & seafood
Shellfish and finfish grading and sorting, defect detection, harvest logging, feed tracking and FSMA 204 lot registers.
Agriculture & produce
Grading and defect sorting, harvest-ready detection, yield forecasting, early disease and pest identification.
Warehousing & logistics
Inventory and pallet tracking, dock-door event capture, cold-chain hand-offs and safety compliance.
Retail
Shelf and stock analytics, footfall and dwell measurement, loss prevention and queue monitoring.
Healthcare
Imaging-analysis support, patient-area monitoring, hygiene and protocol compliance.
Why Tangent
Boutique attention, industrial reliability
We build, not resell
Models trained on your data and your conditions. No repackaged generic detector that half-works on your product.
Hardware-literate
Axis Technology Partner. We speak lens choice, lux, mounting, PoE budget and network load — not just Python.
Compliance-aware
FSMA 204 Key Data Elements, GDST and EPCIS event structures designed in from the start, not bolted on later.
You talk to the builder
A small practice by choice. The person who scoped your system is the person who writes it and answers the phone.
Common questions
Straight answers
How is automated visual inspection different from manual QC?
A person spot-checks a sample and judges it by eye, so results vary with fatigue and who's on shift. An automated system inspects 100% of units against one fixed standard, flags defects in real time, and keeps an image of every unit — so nothing ships unseen and you have evidence of what passed.
What kinds of defects can it catch?
It depends on what a camera can see, but commonly: surface defects (scratches, cracks, discoloration), shape and dimensional errors, missing or misplaced components, label and print errors, foreign objects and contamination, and grade or size banding. We train the model on examples of your good and bad units, so it learns your definition of a defect — not a generic one.
What is passive traceability?
Passive traceability means the record is created by observing work that is already happening — cameras and sensors log the event automatically — rather than by asking a worker to stop and type it in. The compliance record becomes a by-product of the operation instead of an extra task.
Do I need to replace my existing cameras?
Usually not. We're an Axis Technology Partner and work with most ONVIF-compliant IP cameras. Where existing coverage is adequate we build on it; where it isn't, we specify only the additional views the traceability model actually requires.
Does installing a camera system make me FSMA 204 compliant?
No software makes you compliant on its own — compliance is a programme covering your records, your procedures and your traceability plan. What our systems do is capture and structure the Key Data Elements so that producing an FDA Electronic Sortable Spreadsheet within 24 hours is a single export rather than a week of searching.
How long does a deployment take?
A scoped pilot on one line or one site typically runs four to eight weeks from site survey to a working capture-and-export loop. Full rollout depends on site count and network readiness.
Do you work outside food and aquaculture?
Yes. The same detection-and-event architecture runs in manufacturing QC, warehouse and dock operations, retail analytics, and healthcare monitoring. Food traceability is where the regulatory pressure is sharpest, which is why it leads — but it isn't the limit of the work.
Let's look at your line.
Tell us where quality escapes or the line slows down. We'll tell you honestly whether cameras can fix it — and what it would take.