How Does Automated Steel Quantity Extraction Work?
Automated steel quantity extraction reads PDF construction drawings and identifies every structural steel member label — section sizes like W12X26, HSS6X4X1/4, or C10X15.3 — along with their exact location on each page. The best implementations have to cope with every kind of drawing, because no single way of reading a page works reliably across all of them. Here is what the process looks like, using Steelflo as a concrete example.
Step 1: Reading the Drawing
The first job is reading everything on every page of the PDF. Most modern structural drawings produced by Revit, AutoCAD, or Tekla carry a real text layer, and reading that directly is faster and more accurate than optical character recognition. Every label comes back with its exact position on the page.
On these pages, the text layer alone accounts for 80-90% of steel labels, and the positions are precise because they come straight from the PDF rather than from image processing.
Step 2: Separating Steel From Everything Else
Reading the page returns everything on it: title blocks, grid labels, notes, dimensions, revision clouds, door schedules. The system has to pick the steel member designations out of all that noise.
Steelflo recognizes structural steel nomenclature: W-shapes (W12X26), wide-flange columns, HSS sections, pipes, channels, angles, plates, and specialty items like lintels. Each detection is classified by category (beam, column, HSS, channel, angle, plate) and given a confidence level.
It also handles more than one standard. Steelflo automatically detects whether a drawing set uses AISC (US), BS/IS (UK/India/Middle East), AS/NZS (Australia/New Zealand), or EN (Europe) notation, then recognizes designations written in that standard's conventions, like UC305x305x158 or SHS220x220x6.0 for BS/IS and 310UB40.4 or 150PFC for AS/NZS.
Step 3: Pages Without Usable Text
Some pages defeat text reading entirely. CAD drawings with vector fonts, hand-lettered annotations, or embedded raster images have no usable text layer. These pages are common in older drawing sets and certain CAD workflows.
Steelflo spots these pages on its own and reads the steel callouts from the page image instead, with a location for each detection. This is slower and slightly less precise than reading the text layer, but it catches labels that would otherwise be completely missed.
What the Numbers Look Like
On a 7-page US commercial structural package, Steelflo found 53 individual steel label occurrences across 18 distinct section types. A human estimator counted 17 types on the same drawings. Steelflo found all 17, and it also surfaced a flush-beam callout on the plan (W12X17) that the manual count had skipped. That label needed an RFI, because W12X17 isn't a standard AISC size.
On larger international projects, the results scale: 1,047 labels from a convention center drawing set using BS/IS standards (India), and 237 labels from Australian commercial drawings using AS/NZS notation.
From Detections to a Bill of Materials
Each detection is an individual bounding box — one label on one page. The same member type might appear 12 times across plan views, sections, and details. The system aggregates these into a bill of materials: group by member type, sum the count, and look up unit weights from a validated catalog (Steelflo's covers 2,500+ standard sections).
The critical step between detection and final BOM is human verification. Every detection is linked to its source page with a visual overlay on the actual PDF. The estimator reviews each one, confirms legitimate detections, and rejects false positives (a grid label that looked like a steel callout, or the same member shown in both plan and elevation). Low-confidence extractions are flagged automatically so the estimator knows where to focus attention.
Why Handling Every Page Type Matters
A system that can only read the text layer misses vector-font pages entirely. A system that reads everything as an image is slower, more expensive, and less precise on pages where the text is right there. Handling both is what gives you speed, accuracy, and coverage across the full range of drawing quality that fabricators encounter in practice. For the accuracy data behind these results, see how accurate is AI for steel takeoffs. To see it in action, try SteelFlo on your own drawings.