Computer VisionField engagement

The product was not answering which pixels should change. It was proving which pixels must never change, on every image.

Architecture and delivery lead.
zero
unexpected changes, per image
8
implicit judgments made explicit
every change
gated by the harness
candidate detection
constraint enforcement
regression harness
Where the work actually sits. Detection is the visible part and the smallest part. The constraint enforcement and the regression harness are what turn a detector into a guarantee.
What was at stake

A repetitive human visual judgment, expanding, removing, or modifying specific pixel regions by eye, needed to become a reliable automated system. A false positive, a pixel changed that should have been left alone, is treated as more serious than a hard image sent to a person for review. Reliability is the top priority, above throughput and above coverage.

The constraint

The stated ask is automate the edit, and the reflex is to add a model or grow a mask by a fixed amount. The real problem is the negative guarantee: proving which pixels must never change, on every image. That is a different question from which pixels should change, and no amount of detector accuracy answers it, because a detector describes what it found rather than what it promises not to touch.

The fork

The reflex, and the fix.

Road not taken

Add a model and dilate the mask

Pull

Fast to demonstrate, and it looks correct on most images.

Why not

A fixed dilation crosses protected boundaries, and a model alone can only report confidence. Neither can promise that a protected pixel was untouched.

Road taken

Separate detection from modification, then prove the negative

Accepted

A dual-mask architecture, boundary-aware expansion, deterministic rules over the model output, and a pixel-level regression harness as part of the product.

Bought

A demonstrable zero unexpected changes on protected pixels, per image, with hard cases routed to a person.

Decision

Prove the negative, because that is what the operator was actually guaranteeing by eye.

How it was built
01Candidate mask
02Protected mask
03Decision mask
04Bounded expansion
05Regression gate
01

Surface versus hidden reality

The identify then modify then preserve chain explodes into the eight implicit judgments an operator makes at once. Detection is separated from modification by a dual-mask architecture: a candidate mask and a protected mask feeding a decision mask.

02

Expansion that cannot cross a boundary

Boundary-aware expansion grows the region where it is safe and clips cleanly against an explicit protected mask, with the guarantee that the count of unexpected changes is zero, demonstrated per image rather than asserted.

03

A hybrid pipeline instead of blindly adding a model

The model is demoted to a candidate detector, deterministic rules enforce the exact pixel constraints, and validation decides safety. The model earns its place through measured reliability rather than a demonstration.

04

Integration designed around the engine

Four integration architectures were considered, native, plugin plus engine, hybrid, and headless, and the choice was made for reliability and testability rather than for convenience.

05

A pixel-level regression harness

An automated release gate: build, run the corpus, diff, check protected and expected and performance, pass or fail. Plus a closed production feedback loop from edge case to root cause to corpus to fix to verify.

How it was measured

A single headline number hides where a system fails. This work was scored on the dimensions that actually decide whether it holds in production, measured on real, held-out cases rather than the demo path.

Unexpected changes on protected pixelsReview rate on hard casesRegression pass rate per releasePerformance against the corpus
figures

What it produces
Without this discipline

A system that is right most of the time, cannot say which pixels it promises not to touch, and puts the burden of catching its mistakes back on the person it was supposed to relieve.

This system

A pipeline that guarantees zero unexpected changes on protected pixels and demonstrates it per image, routes hard cases to a person, and gates every change through a regression harness that is part of the product rather than an afterthought.

dual-mask architectureboundary-aware expansionmodel as candidate detectorharness as product
The operating envelope

What it owns, and what it hands to a person.

Handled with confidence
Defined protected regions
Cases inside the corpus
Hard cases routed to review
Flagged for review
Low-confidence candidates
Edge cases not yet in the corpus
Out of scope by design
Guarantees about detection recall
Unattended operation on novel image classes
The honest limit

The guarantee is about protected pixels, not about perfect detection. Hard images are routed to a person by design, and the corpus has to grow as new edge cases arrive, which is why the feedback loop is part of the architecture rather than a process document.

What it generalizes to

Where the cost of a wrong action exceeds the cost of asking a person, the deliverable is the negative guarantee rather than the positive capability. Separate detection from modification, enforce the constraint deterministically, and make the regression harness part of what ships.

How we engage

You have a system like this one.
Tell us where it stands.

Whether it is failing, not yet built, or about to meet a scale it has never seen, we can tell you what we see.

Start a conversation
mostafa@opulion.dev · Response within 24 hours · By inquiry