Context
A fourth-generation, family-owned tannery in Santa Croce sull'Arno — the Arno valley district between Pisa and Florence that is Europe's most concentrated vegetable-tanning cluster, with some 200 to 250 tanneries employing several thousand people. The client employed around 60 staff and supplied finished leather to mid-tier Italian fashion and furniture brands, with a strong quality reputation but a grading process carried out entirely by manual visual inspection: several minutes per hide, dependent on a handful of experienced graders whose judgement was difficult to standardise or document. Three pressures were converging at once. Larger fashion houses were asking suppliers for objective, per-hide defect data — mapped defect locations, cuttable-area percentages, consistent criteria — rather than a grader's subjective sign-off, which the client could not produce without significant manual record-keeping. Manual grading was erring on the side of caution around ambiguous defects, and inconsistency between shifts and between individual graders was causing avoidable downgrading of hides — a direct margin loss on a high-value material. And the most experienced graders were nearing retirement, with no confidence that newer staff could be trained to the same standard in time. The client had examined the machine-vision grading systems already running at larger tanneries in the district, but the specialist providers serving this niche — mostly US and Northern European — priced hardware and service contracts for an export-scale tannery, not a 60-person family business.
Our Mission
Identify a machine-vision hide-grading system that would deliver the objective, per-hide defect documentation the client's brand customers were requesting, at a capital and support cost a 60-person family tannery could justify — and take the client from technical specification through supplier qualification, contracting, installation and operational handover without asking the owners to gamble their quality reputation on an unproven machine.
Our Approach
MAXAM began with technical requirements translation, working with the client's production manager to define what 'grading' actually meant in operational terms — the specific defect types (scars, insect bites, grain damage, brand marks), the imaging resolution needed to catch them reliably, and the reporting format the client's key brand customers were genuinely requesting — before evaluating any equipment. Supplier sourcing then ran through MAXAM's network in China, identifying machine-vision manufacturers with proven high-resolution surface-defect inspection experience in materials-adjacent applications: general industrial surface inspection, textile inspection, and leather-specific AI defect classification systems developed by Chinese research groups and equipment makers serving China's own large domestic leather and footwear industry. Three candidates were shortlisted and required to run trial inspections on hide samples supplied by the client, benchmarked against the gradings of the tannery's own senior grader. MAXAM then negotiated a contract structure proportionate to the client's size — a smaller upfront hardware cost paired with a subscription-style software and support agreement, rather than the large one-time capital outlay quoted by Western providers — and secured remote software support with defined response times, given the tannery would have no in-house machine-vision expertise. Finally, MAXAM supported physical integration of the scanning line into the existing finishing process and, critically for a family business wary of replacing experienced judgement with a machine, structured a parallel-running transition period in which the system's grading was compared against the senior grader's manual assessment, so the owners could build trust in its consistency before relying on it for customer-facing documentation. The wider point holds beyond leather: the underlying problem — grading a naturally variable material against defect classes at a cost a mid-sized family business can justify — is the same one MAXAM has documented in Portugal's cork industry, and a system proven for one materials-grading application is frequently adaptable to another with the right calibration and validation work.
Before & After Results
| Metric | Before | After |
|---|---|---|
| Grading time per hide | Several minutes (manual) | Under 30 seconds |
| Per-hide defect documentation | None (subjective sign-off) | Stored defect map for every hide |
| Capital outlay | Western provider quotes | Meaningfully lower — viable at this scale |
| Senior grader role | Full-time manual grading | Supervision and exception handling |
"We were not looking to replace our grader — we were looking to make sure his judgement survived his retirement. Running the system alongside him for months was what convinced us. By the time we put its output in front of a customer, we already knew where it agreed with him and where it did not."
Owner and Managing Director
Owner and Managing Director — Tuscan Family Tannery