ProductionVerified Real deployment at a named or precisely described company, with a measured result from a primary source.

AI vision grades every board and picks the cut at the edger and trimmer, so more value comes out of each log (Waipapa Pine, Tenon, Hunt Forest Products)

Added 2026-11-16 · Last updated 2026-11-16

Manufacturers in this case: Waipapa Pine · Tenon · Hunt Forest Products · all manufacturers using AI

Role

Wood products production (sawmill, veneer and plywood lines) · Typical users: sawmill manager, production manager, optimization or process engineer, quality and grading supervisor, plant manager of a wood products, furniture or pallet plant

The problem

In a sawmill or veneer mill, logs are the biggest cost, and the value of each board depends on decisions made in seconds: where to edge, where to trim, which grade to give, what to cut out. When people make these decisions by eye, or older scanners miss defects, good wood is wasted and lower grades are produced. Waipapa Pine (Kerikeri, New Zealand), which saws radiata pine into framing timber, faced rising log prices driven by export demand and an expected drop in the regional harvest, and its board edger was manually fed at 8 to 10 boards a minute. Tenon's Clearwood plant (Taupo, New Zealand), one of the country's largest producers of defect-free appearance grade radiata pine, buys pruned logs on the open market and cannot afford to waste them. At Hunt Forest Products (Pollock, Louisiana, USA), a veneer and plywood plant of about 290 people, light-coloured wane on green veneer was not being picked out, so those pieces took up dryer space and were downgraded later at the dry graders.

How AI helps

Vision sensors scan each board or veneer sheet, a deep learning model trained on scans of the mill's own wood finds and classifies the defects, and an optimizer chooses the edging, trimming, clipping and grade that give the highest value for the products the mill can sell. At Waipapa Pine, USNR's BioVision system at the edger uses deep learning for defect recognition and classification, and a transverse high grader with four-sided and end-grain scanning grades both green and dry boards on a combined trim line. At Tenon, sensors above and below the board look for knots, pitch and bark pockets, latewood resin, bird's eye, harvest damage and pith; the mill supplied enough board scans showing each defect to pre-train the deep learning networks before start-up, and the edger and trimmer share the same logic, so lower-grade boards can be sent back to the edger to recover high-value wood. At Hunt Forest Products, deep learning was added to the existing USNR NV4g green veneer scanner so wane is found and clipped at the green end, before drying.

Step by step

  1. Measure today's baseline per line: log volume in, lumber or veneer out (recovery), grade mix, output per shift, and the downgrades found later in the process.
  2. Find where decisions lose the most value: edger, trimmer, green or dry grading, or rework in secondary processing, using audits of boards pulled from the line.
  3. Ask suppliers whether your existing scanners can get a deep learning upgrade before replacing the line.
  4. Plan the training data with the supplier: collect scans of your own species and defects before start-up, labelled by your best graders.
  5. Keep the same before and after measures and separate the effect of the AI from mechanical changes made in the same project.
  6. After start-up, have graders audit samples every week and feed corrections back so the model keeps learning.

Tools

  • USNR BioVision edger optimization with deep learning defect recognition and a Transverse High Grader: used at Waipapa Pine
  • USNR BioVision edger and trimmer optimization with BioLuma 2900LVG sensors and deep learning: used at Tenon Clearwood
  • USNR NV4g green veneer scanner with deep learning: used at Hunt Forest Products
  • Other tools to evaluate: grading and optimization scanners from other sawmill and wood processing equipment suppliers, AI upgrades for your existing scanners, vision inspection for cut-up lines in furniture and pallet plants
  • To start without new equipment: your recovery and grade reports and any LLM (ChatGPT, Claude, Copilot, Gemini) with the prompt below

In the AI tools index: USNR BioVision · USNR NV4g veneer scanner · all AI tools used by manufacturers

Result

Waipapa Pine (USNR article, the equipment supplier; project completed early 2020): 4.2% better sawn recovery, 8.3% better grade recovery and 13% more output, with 95% of operational jobs in the mill now hands off and manual graders retrained on the new technology. Tenon Clearwood (USNR article): overall piece count up 20%, though not a 20% volume increase, because accurate grading at the green mill lets it cut dimensional pieces there instead of sending wide random boards to secondary processing; the mill reports better throughput, more consistent grading and better conversion. Hunt Forest Products (USNR article): much less downgrade at the dry graders since installing deep learning, and the mill manager said the system keeps them from wasting dryer space on veneer that would become waste; block count rose by close to 1,000 blocks a day after the same project, which also replaced a bottleneck diverter. USNR said 37 systems with this deep learning technology had been sold at the time of the Tenon article.

As reported by the source; most figures are vendor or self-reported and not independently verified.

Human review: cautions

  • All results come from the equipment supplier (USNR) and quote the mills; they are not independently audited.
  • Each project also changed machinery (new edger, trim line, high grader or diverter), so the gains cannot be credited to AI alone.
  • Tenon's 20% piece count gain is not a 20% volume gain, as the mill points out.
  • Deep learning needs a large set of scans of your own species and defects; plan time for training and grader audits before start-up.

Try it yourself

You are a sawmill optimization engineer. I want to know where better grading and cutting decisions would add the most value in our mill.

I will paste: (1) monthly production reports for the last year [log volume in, lumber or veneer volume out, grade mix, output per shift, downtime]; (2) results of board audits or downgrade reports [station, defect type, grade given, correct grade, value lost if known]; (3) the products we can sell with their price per grade.

Do this:
1. Calculate recovery and value per log volume by month, and show the trend.
2. Rank the stations (edger, trimmer, green grading, dry grading, secondary processing) by value lost, with the main defect types at each.
3. Estimate the value of a 1% gain in recovery and a 1% gain in grade mix with our prices, and say which assumptions this depends on.
4. List what a scanner and AI supplier would need from us (species, defect list, training scans, line speeds, product list) and the questions to ask them.
5. Draft a before and after measurement plan that separates the AI effect from any mechanical changes.

Use only the data I paste in; if something is missing, write 'not found' or list it as a question instead of guessing. Do not send, order, post or change anything in any system: produce drafts for my review and approval.

Want more like this? Join the newsletter

Related cases

All 21 Production use cases →

Sources

Weekly: 1–2 practical AI use cases for your plant

Newsletter in

Free, one email a week. By subscribing you agree to receive the weekly newsletter (and its sponsor message) from AI Agents for Manufacturing. Every email includes an unsubscribe link, and you can unsubscribe anytime. See our Privacy Policy.