Frozen Vegetable Defect Specifications: Colour, Peel and Insect Limits
Jan 16, 2026
Leave a message

Research This Article & Topic with AI
Verify key claims, compare reliable sources, and explore GreenLand-food's practical frozen food knowledge.
Article details and research questions are prepared in the links. Sign-in or sending may be required.
Reviewed 16 October 2026. A useful frozen-vegetable tolerance clause names the exact defect, the product form, the denominator, the sample and the decision rule. "Black skin below 1.5%" cannot settle a claim until buyer and supplier know whether the number counts affected dice, weighs residual peel, or measures dark surface area. A reported local market expectation is a question to document, not evidence of a law.
AUDIO SUMMARY · ENGLISH
Frozen Vegetable Defect Specifications: Colour, Peel and Insect Limits
Listen to the key points from this article. 3:58 · English · AI narration.

A proposed inspection of residual peel on frozen diced water chestnuts.
A buyer recently reported that its Taiwan customer wanted residual black skin on peeled diced water chestnuts below 1.5% after receipt and described the figure as a local government or market requirement. We have not been given a legal citation, official notice, signed product specification, sampling denominator or inspection method for that statement. We would first treat it as the buyer's proposed commercial condition, request the underlying document, and preserve the receiving evidence. The same approach is useful for broccoli color, insect findings and other frozen vegetable claims: identify what was seen, count or weigh it consistently, then apply the rule actually agreed for the shipment.
Start with the actual defect and buyer claim
The product in the reported case was peeled, diced, individually quick frozen water chestnut for industrial use. Its cut form matters. A dark sliver on the outside of one white die may be residual external skin left after peeling; a brown mark in the exposed tissue may be bruising, oxidation or a different quality defect; blackening through the center may indicate internal blackheart or damaged raw material. A receiving inspector who calls all three "black skin" changes the numerator without warning. We would ask for close photographs of several affected pieces, photographs of their opposite faces and cut interiors, and a retained frozen sample before proposing a remedy.
We would also obtain the purchase order, signed specification, approved pre-shipment sample, artwork if there is a customer-facing pack, and the buyer's exact post-receipt report. Were the dice assessed frozen under controlled light, thawed after a set time, or heated in the actual recipe? How many cartons were opened, from which pallets, and which lot codes did they carry? Was the finding isolated to one bag or distributed across the container? These questions determine whether the issue is a peeling setting, raw-material selection, rework opportunity, storage damage or simply a different visual definition. They do not ask the customer to complete an elaborate form; GreenLand can gather the few missing facts in a working conversation.

Actual GreenLand water-chestnut product photograph relevant to this defect definition.
The buyer's statement that a Taiwan authority or market requires "under 1.5%" needs a source. A regulation would require its issuing body, effective date, commodity scope, legal wording and prescribed method. A retailer specification would require a controlled document and version. A customer's internal preference can still be commercially important, but it should enter the next order as an agreed clause, not be presented as law to another customer. Until those details are supplied, we should not say that Taiwan legally permits, requires or prohibits a particular percentage of water-chestnut skin. Our frozen water chestnut slices product page shows the adjacent peeled form and describes peeling, trimming and spot control; the dice in this claim still need their own signed form specification.
Start the record with two columns in prose: what the buyer observed and what the parties had agreed before shipment. Record any unagreed threshold as a proposed change for the next production plan. For the current lot, GreenLand would compare the retained origin sample and documented inspection method with the receiving sample, then discuss sorting, conditional acceptance, rework or claim resolution according to the contract and the actual extent. This preserves the buyer's concern without inventing a government standard or dismissing a visible defect.
Separate peel, surface spots and internal defects
Residual peel is material from the water chestnut's external dark covering attached to an otherwise peeled piece. A test needs an operational boundary: for example, any visible dark external skin attached to a die after loose fragments are removed. The inspector should not count a detached speck twice, once as peel and again as a defective die. If the contract uses a count basis, one affected die is one defective unit regardless of the small peel area; if it uses a peel-mass basis, the inspector separates and weighs the actual peel. These are different measurements with different commercial meanings.
A dark surface spot may be attached peel, oxidation, bruising, physiological discoloration or localized decay. A photograph under stable illumination can show size and contrast, but it rarely proves cause by itself. We would compare the spot with raw-material and peel-trim records and, where appropriate, cut several pieces to check whether discoloration extends below the surface. A buyer may choose a maximum visible area on a piece, a count of affected pieces, or graded categories such as minor and severe blemish. The specification must explain whether multiple small marks on one die are aggregated, how borderline colors are judged and whether both faces are inspected. A marked reference image can be useful, provided it is tied to the written rule.

Actual GreenLand water-chestnut product photograph relevant to this defect definition.
Internal blackheart is different again. It can be invisible in a whole peeled water chestnut and obvious only after cutting. An optical sorter viewing the exterior cannot prove that every interior is sound. A diced product exposes many interiors, so raw-material selection and cut-stage checks are relevant. If the customer calls an internal dark center "residual black skin," the factory may chase a peeling adjustment that cannot fix it. We would separate those findings on the inspection sheet and use their frequency by raw-material batch to decide whether to strengthen incoming selection or the cut-stage removal step. A dark cut face that grows during storage may require a process and temperature investigation as well as visual grading.
Water chestnut also carries product-specific nonconformities that should not be folded into the black-skin numerator: root or mud residue, hard field matter, oversized or undersized dice, soft texture after cooking, fused clumps, broken pieces and excess fines. A private Foodnet water-chestnut specification expressly separates adhering skin from spots, blemishes and foreign material. A Frucom sliced-water-chestnut specification likewise distinguishes blemish sizes, broken pieces and physical hazards. Those are examples of buyers writing separate categories, not universal limits to copy into a Taiwan dice order.

Three different physical locations of a dark finding require separate defect definitions.
For each category we would agree whether it is cosmetic, functional or safety-critical. A small peel fleck may be a buyer appearance failure even where food safety is not implicated. Glass, metal and hard plastic call for a separate hold and investigation, even if their measured percentage is tiny. A suspected insect or plant fragment must be identified before it is assigned to a legal or buyer category. We would make a photo library of agreed pass, borderline and fail pieces from the approved sample, then use it to train the origin and receiving inspectors against the same definitions.
Write the denominator and strict limit into the clause
"Less than 1.5%" means a strict inequality: a measured result of exactly 1.5% fails. It is not the same as "no more than 1.5%." If the method counts pieces, the calculation is 100 × affected pieces ÷ all examined pieces. In a 200-piece sample, three pieces with residual peel equal 1.5% and fail a strict <1.5% clause; two equal 1.0% and pass. In a 1,000-piece sample, 14 affected pieces equal 1.4% and pass, while 15 equal 1.5% and fail. The parties should not round 1.49% to 1.5% first and then decide. The clause should specify the number of decimal places reported and compare the unrounded calculation with the limit.
A mass basis asks a different question: how much of the sampled product is peel? If a 1,000 g sample yields 15 g of separated residual external peel, the result is 1.5% by mass and fails a strict <1.5% clause. A result of 14 g is 1.4% and passes. This method needs a written separation protocol, a scale resolution appropriate to the limit, a decision on adherent versus detached peel and a rule for moisture or surface ice. Weighing entire defective dice as the numerator would not measure peel mass; it would measure mass of affected product. Both can be valid contractual measures, but they cannot share one unexplained "% skin" label.
Area basis is useful if the customer cares about what appears on a white die. The numerator could be the total visible dark external-peel area and the denominator the total visible face area inspected under a defined camera geometry. That is more demanding than counting affected dice. The fixture, lighting, camera distance, color threshold, number of viewed faces, occlusion rule and manual adjudication of borderline spots all influence the result. A photograph of one side cannot establish an all-surface percentage. We would not turn an area threshold into a laboratory mass result or declare a product compliant based on a casual phone picture.

Exactly 1.5% fails a strict below-1.5% clause; counting affected dice and weighing removed peel measure different quantities.
The contract must also tell the inspector what sample is being evaluated: one bag, a composite across selected cartons, one production lot, or an entire shipment. We would write the product form, defect definition, inspection condition, denominator, sample quantity, collection method, formula, precision, threshold sign and disposition in one clause. An example for negotiation is: "On a frozen, randomly selected 1,000-piece sample of IQF peeled water-chestnut dice from the defined production lot, count each die with attached visible external dark peel once; residual-peel affected-piece percentage = count ÷ 1,000 × 100, with an acceptance limit strictly below 1.5%, calculated before display rounding." That wording remains a proposed buyer specification until signed. It does not address internal blackheart or surface oxidation, which need separate clauses.

One affected die can generate different count, mass and area percentages.
For larger or irregular dice, a buyer may prefer a mass-based sample; for a retail-facing ingredient, affected-piece count may track consumer perception better. The parties can trial both methods on the same retained samples and choose the one that predicts actual application complaints. We would test repeatability between two inspectors before production, because a mathematically exact threshold is not helpful when the defect definition changes with the person holding the tray.
Build a defensible lot sample and photo record
Sampling starts with a lot boundary. One container can contain several production days, cut settings or raw-material intakes. If the reported defect is concentrated in one sublot, pooling everything into one average may conceal it; if the concern is a single opened carton, treating that carton as the entire shipment may exaggerate it. We would map carton marks to processing date, raw-material group, peeling line or approved processor, and pallet location. The origin and receiving teams should inspect the same unit definition and sample frame, with the planned locations and number of cartons agreed before either party sees the result.
For a prospective trial, we would select cartons across pallet positions and production codes, then draw the agreed number of bags and dice without searching only for clean or visibly poor material. Keep targeted samples from a visibly affected carton separate from the random acceptance sample and label both. Record carton and bag IDs, frozen condition, inspection time, who selected the units, scale or camera used, and the exact classification. A second inspector should classify a subset independently; large disagreement is evidence that the definition or reference images need work. Retain sealed split samples at suitable frozen conditions if the buyer and supplier may need an independent recheck.
Photo evidence should be useful to a person who was not present. Photograph the unopened carton and lot code, the opened bag, the full inspection tray, each representative defect beside a scale, the reverse or cut face where relevant, and a clean comparator under the same light. Do not crop away the sample ID. Store the original file and record any later annotation separately. A handful of close-ups proves that a defect exists but does not by itself establish its percentage in the lot. The full-tray and denominator record connect the close-up to the claim.

An agreed random sample covers identified pallet positions and production codes; suspect cartons stay separately identified.
The current ISO 2859-1:2026 provides AQL-indexed acceptance sampling schemes for lot-by-lot attribute inspection. It asks the parties to choose the lot size, inspection level, scheme and acceptance/rejection numbers. An AQL of 1.5 is a sampling-plan index describing a process average under the standard's rules; it is not an assurance that every carton or every lot has less than 1.5% residual skin. Nor can an AQL table decide a mass or area percentage unless the parties define suitable attribute units and decision rules. We would write the version and plan into the contract, then record the actual code letter, sample size, accept number and reject number for the lot before inspection.

Carton identity, full tray result and frozen split support independent rechecks.
Retesting must be agreed before a dispute. A sensible path is to freeze and identify the first sample, notify both parties, use an independent inspector on retained or newly selected units according to a stated trigger, and keep the first result visible rather than replacing it quietly. A targeted investigation may show a bad sublot even when the original random lot sample passed. Conversely, one visually poor piece does not prove a statistical lot failure. Our quality inspection and sampling guide covers the broader shipment file; here the issue is the exact defect numerator and denominator.
Use commodity standards without copying their numbers
Commodity standards are useful because they define product forms and observation methods. They do not make one buyer's cosmetic limit universal. The USDA frozen broccoli grade standard treats broccoli styles differently: its listed sample unit for short spears or florets is 50 units, while cut and chopped styles use 250 g units in the AMS tally sheets. The standard also distinguishes a blemished unit from extraneous vegetable material and uses its own AQL terminology. That structure is a useful reminder to identify the form before choosing a method. A broccoli floret count cannot be transplanted to peeled water-chestnut dice or to a 1,000 g peel-mass sample without a technical decision and a signed buyer clause.
Insect findings need equally careful reading. The FDA Food Defect Levels Handbook lists an action level for frozen broccoli averaging 60 or more aphids, thrips or mites per 100 g under the specified method; the FDA broccoli compliance policy guide describes that number as a criterion for direct-reference enforcement action. It is not an instruction to accept 59 insects, a promise that lower levels meet every law, or a buyer's quality target for a premium pack. Other contaminants, food-safety conditions and the customer's contract can trigger a hold below that number. Its commodity is broccoli, not water chestnut.
The insect count also should not be merged with "black spots." A tiny attached dark mark on a water-chestnut die might be peel; a fragment suspected to be insect matter requires identification and a different evidence path. We would isolate the object, preserve the sample, review upstream field and wash controls, and involve a competent laboratory when identification matters. A visual sorter can reject a conspicuous dark foreign object but cannot replace the identification, safety assessment or trend review. A buyer asking for an insect-free appearance may contract for a stricter target than a regulatory action level; GreenLand would confirm whether that target can be supported by the actual source and inspection method before quoting.

USDA grading units and FDA insect enforcement criteria serve different purposes and apply to broccoli.
Private specifications show why percentages need names. One published water-chestnut document uses "adhering skin" as a separate weight-based line; another grades dark blemishes by size and counts per 1,000 g. The fact that two professional buyers use different designs is evidence against treating their limits as a government number. For our Taiwan case, we would ask the buyer for its version-controlled local customer document and compare it with our approved reference sample. If an official legal requirement is supplied, we would identify the authority, exact product category and test method before updating the market file. Until then, the <1.5% figure belongs only in the negotiated contract conversation.
Match sorting and blanching controls to the defect
An action plan should follow the cause and where it can be observed. For residual external peel, we would confirm incoming water-chestnut skin condition, peeling method and settings for the actual size range, rework rules, trim inspection and the cut-stage visual check. We would compare a small controlled trial before and after any setting change for edible yield, attached peel, broken pieces and texture. An aggressive peel that improves one percentage can increase waste and damage the white flesh. The buyer needs the combined result. If the issue is internal blackheart, we would focus on raw-material selection and exposed cut faces rather than merely speeding up the peeler.
Visible color and attached peel are optical inspection problems. An RGB camera can use controlled visible-light color and shape contrast to flag dark areas on white water-chestnut surfaces. A near-infrared or multispectral channel may help distinguish materials under the equipment and calibration actually used. Both are limited by occlusion, overlapping pieces, moisture, lighting and the training set; a sorter needs a thin, separated product presentation and validated challenge pieces for the exact SKU. We would describe such a control as proposed until the partner site confirms its installed equipment, sensitivity, line speed and reject checks. Human review of the reject stream and a post-sort sample still matter.
X-ray inspection answers another question. It detects differences in X-ray absorption or density and can be useful for certain dense foreign materials, depending on the product and machine settings. It is not a color camera and should not be credited as the control for black external peel or brown surface spots. Metal detection likewise targets detectable metal, not every cosmetic defect. A procurement specification can name the desired detection outcome and validation pieces without claiming that GreenLand or a particular plant already operates a device that has not been verified for that order. This prevents a reassuring equipment name from concealing an unaddressed peel problem.

Visible peel calls for optical contrast and cut-face checks; X-ray addresses suitable density contrasts in foreign matter.
Blanching may influence enzyme activity, color stability and texture in some vegetables, but its result depends on product form, time, temperature and cooling. The University of Minnesota Extension explains the quality purpose of blanching and that freezing does not kill bacteria present on vegetables. Water chestnut's crisp bite can be changed by heat, so we would trial a blanched and an unblanched version against the customer's cooked application before proposing that step. Blanching cannot remove attached black peel. It should not be called a validated pathogen kill step unless the specific process has been validated for that purpose. Packaging, quick freezing and stable frozen transport preserve the selected quality; they cannot correct a misclassified defect after packing.
Decide pre-shipment acceptance and post-arrival claims
Before the next order, we would prepare a short, usable specification annex with the buyer. It would state peeled water-chestnut dice size and form, external skin definition, separate black-spot and internal-defect rules, agreed illumination and sample condition, denominator, strict or inclusive inequality, lot boundary, random sampling route, precision, retest path and intended application. It would also cover critical foreign matter and any functional quality requirement such as bite after the buyer's cooking process. If the customer prefers an approved physical sample, preserve it under an identified condition and use it alongside the written rule, not as a substitute for the arithmetic.
Our proposed origin sequence would put the same rule at each handoff:
- Record the raw-material batch, incoming condition, size range and peel settings for the planned trial.
- Compare the changed settings with the current method for residual peel, edible yield, breakage and texture, then classify exposed cut faces separately.
- Inspect the IQF lot using the signed numerator, denominator and sample route; preserve representative photographs and any agreed retained split.
- Link the accepted result to carton codes, packing checks and the shipment file, then send the buyer the raw count or mass figures with the pre-shipment evidence.

Compare the receiving finding with the signed specification, preserving both the raw result and the agreed decision rule.
We coordinate the inspection or laboratory work needed for the agreed order and connect each result to the relevant lot. A named GreenLand contact follows the specification, shipment records and receiving questions through to a documented decision.
At arrival, the importer or receiving QA team should first preserve the state of the shipment: reefer and temperature data if available, carton marks, seal condition, packaging damage, frost, clumping and pallet map. Then it should draw the pre-agreed sample and classify the defects while frozen or under the agreed controlled preparation. If a black piece appears, photograph and retain it before sorting the rest of the tray. The report should show both the raw numerator and denominator-such as 3/200 or 15 g/1,000 g-rather than only a rounded percentage. That allows both teams to reproduce the calculation and decide whether a strict "less than" term failed.
If the current purchase did not contain the reported 1.5% term, we would investigate the observation and commercial impact without retroactively presenting the new threshold as an existing legal duty. The practical resolution might be additional sorting, a controlled trial of another cut or peel setting, price adjustment, replacement supply or rejection under the actual contract and applicable law. We would keep the first sample, any retest, the buyer's reference document and the decision trail together. A corrective action should target the confirmed defect mechanism and be tested on the next production lot; a generic promise of "better quality" cannot be audited.
The same discipline extends to frozen broccoli color and insect specifications, where product form and method determine the meaning of a number. GreenLand can help the buyer turn a verbal concern into a short accepted rule, test it on a real sample, and carry it from production to receiving. That is the strongest basis for a repeat order because both sides can see what was measured and what action follows when the measured result crosses the agreed limit.
Related reading
Set the broader lot, arrival and retest process around a written defect clause.
Separate ice and net-weight calculations from peel and visual defect percentages.
Agree a measurable frozen vegetable specification
GreenLand-food is a frozen vegetables supplier and manufacturer in China discussing factory-direct wholesale supply. Tell us the product form, the visual or physical issue, pack, quantity, application, destination and available reference sample. We can compare a workable inspection method and order specification with the actual source.

