Detecting Patchiness with AI Vision in Textile and Apparel Manufacturing
How NorrStudio by NorrSpect identifies unlevel dyeing, patchy colour distribution, and localised shade irregularities in dyed and finished fabrics at production speed eliminating blotchy appearance defects before they reach cutting, garment assembly, or retail.
93%
Reduction in patchiness-related roll downgrades and buyer rejections
8cm²
Minimum patchy zone area detectable inline at production speed
97.9%
Detection accuracy across reactive, disperse, and sulphur dye systems on woven and knit substrates
Overview
Patchiness the irregular, blotchy distribution of colour intensity across a dyed fabric surface is one of the most subjective and consequently most disputed defect categories in textile quality control. Unlike a discrete stain or a measurable shade bar, patchiness manifests as a diffuse, graduated variation in dye uptake that may cover large areas of the fabric with no sharply defined boundary. It is difficult to describe objectively, difficult to reproduce consistently in manual inspection, and difficult to communicate to a dye house as an actionable process fault yet it is immediately and unmistakably visible to the end consumer as a fabric that looks uneven, cheap, or poorly dyed.
NorrStudio, developed by NorrSpect, replaces subjective manual patchiness assessment with continuous spatial texture and colour uniformity analysis quantifying the degree of dye unevenness across every square centimetre of the fabric roll and expressing it as an objective uniformity score against the approved fabric standard, eliminating inspector-to-inspector variation and providing the dye house with actionable process data to address the root cause.
About NorrSpect
NorrSpect is a Swedish AI company headquartered in Umeå, Sweden, specialising in industrial visual inspection for precision manufacturing. Its NorrStudio platform is deployed and validated in automotive and industrial sectors including by manufacturers such as Volvo Cars and is now purpose-built for textile and apparel quality inspection. Patchiness detection sensitivity and uniformity scoring models are defined and validated during the pilot phase using the client's approved fabric standards and dye system parameters.
Industry challenge: why patchiness is the most disputed defect in dyed textiles
The core challenge with patchiness detection is its gradient nature. A hole is binary it is either there or it is not. Patchiness is a spectrum: from barely perceptible dye unevenness that a lenient inspector passes and a strict buyer rejects, to severely blotchy fabric that everyone agrees is a write-off. The absence of an objective measurement standard means that patchiness disputes between mills and buyers are resolved by negotiation rather than data an expensive, relationship-damaging process that NorrStudio's objective uniformity scoring eliminates.
The process causes of patchiness are varied and often interacting: inadequate fabric preparation allowing uneven dye site accessibility, insufficient agitation during exhaust dyeing causing localised dye concentration gradients, fabric creasing in the dye bath, uneven chemical auxiliaries distribution, or inconsistent fixation conditions. Each cause produces a visually distinct patchiness pattern and NorrStudio's spatial analysis can distinguish between them, pointing the dye house toward the specific process variable responsible.
Exhaust dyeing unlevel patches
Irregular dye uptake zones caused by inadequate liquor circulation in jet or jigger dyeing producing cloud-like patches of higher and lower colour intensity across the fabric length
Crease mark patchiness
Sharper-edged patches corresponding to fabric fold or crease positions in the dye bath the creased zone experiences different dye accessibility, producing a repeating linear patch pattern
Preparation patchiness
Uneven scouring or bleaching prior to dyeing leaves localised zones of higher or lower fibre absorbency causing a patchy dye uptake that reflects the preparation irregularity rather than a dyeing fault
Auxiliaries distribution patchiness
Uneven application of levelling agents, migration inhibitors, or fixation chemicals creating zones of different dye fixation efficiency visible as soft-edged patches of varying depth across the fabric surface
Pad-batch patchiness
Uneven pad liquor pick-up across the fabric width during pad-batch dyeing caused by mangle pressure variation or pad trough concentration drift producing width-directional patchiness zones
Fixation patchiness (reactive dyes)
Uneven alkali distribution or fixation temperature during reactive dye fixation producing areas of different dye-fibre bond density visible as alternating depth zones on the fabric face after washing off
Solution: NorrStudio AI patchiness detection and uniformity scoring
NorrStudio uses high-resolution multi-channel colour imaging combined with spatial texture uniformity analysis to detect and quantify patchiness across the full fabric width at production speed. Rather than simply flagging a colour anomaly at a point, the system analyses colour intensity variation over defined spatial windows measuring the statistical distribution of colour values within each window and comparing it to the uniformity baseline of the approved fabric standard. Zones where local colour variance exceeds the approved uniformity threshold are classified as patchy and annotated on the roll map with their location, area, and severity score.
Detects patchy zones as small as 8cm² by analysing spatial colour variance within sliding measurement windows across the full roll width
Generates an objective patchiness uniformity score per roll replacing subjective inspector grading with a quantified, repeatable measurement
Distinguishes crease mark patchiness (linear pattern) from exhaust dyeing cloud patches (diffuse gradient) and preparation patchiness (fixed-position zones) by spatial morphology analysis
Identifies pad-batch width-directional patchiness and correlates it to mangle pressure profiles for immediate process correction
Detects fixation patchiness on reactive-dyed fabrics by analysing depth-of-shade variation after washing off the final state visible to the buyer
Provides roll-level uniformity reports with spatial patchiness maps that serve as objective evidence in buyer-mill quality disputes
Integrates patchiness severity data with dye house process logs linking specific uniformity scores to liquor ratio, agitation rate, and fixation temperature for root cause analysisDeployment summary
Solution
NorrStudio AI Inspection Patchiness Detection Module
Inspection scope
Dyed woven and knit fabrics across reactive, disperse, sulphur, and vat dye systems post-washing off
Hardware
High-resolution multi-channel colour line-scan cameras, calibrated uniform illumination, motion-sync encoder
Output
Real-time patchiness alerts, spatial uniformity maps, roll-level uniformity scores, PDF QA reports
Integration
Dye house process logs, ERP / WMS, pad mangle control systems, buyer quality documentation portals
Deployment time
Pilot phase calibrated to client dye system, fabric construction, and buyer uniformity tolerance before full deployment
Use case: reactive dye house patchiness elimination on cotton jersey for mid-market retail
The problem: A reactive dye house processing cotton single jersey in 200–280 GSM weights for mid-market European retail brands was experiencing a persistent patchiness problem on its jet dyeing range — approximately 10–14% of rolls per dye lot were being downgraded or returned by buyers citing unlevel dyeing. The dye house's outgoing inspection relied entirely on inspector visual assessment under a standard D65 lightbox, with no objective uniformity measurement and significant inspector-to-inspector variation in pass/fail decisions. Buyer disputes over borderline rolls were consuming significant management time and eroding the commercial relationship.
The NorrStudio solution: NorrStudio was installed at the stenter exit after washing off. Spatial uniformity analysis was calibrated to the dye house's 80 active reactive shades and the buyers' respective uniformity tolerance thresholds. The system immediately identified that cloud patchiness on the jet range was most severe on rolls where liquor ratio had dropped below 1:6 a process variable not previously correlated with patchiness outcomes. Minimum liquor ratio controls were implemented, and the objective uniformity score replaced inspector visual assessment as the primary outgoing QC criterion.
Results:
Metric | Before NorrStudio | After NorrStudio |
|---|---|---|
Roll downgrade rate from patchiness | 10–14% per dye lot | <1.2% per dye lot |
Buyer disputes over borderline patchiness | 6–10 per quarter | 0–1 per quarter (objective score provided) |
Inspector-to-inspector pass/fail consistency | Variable — subjective assessment | 100% consistent — objective uniformity score |
Root cause process variable identified | Unknown | Liquor ratio below 1:6 — identified in first week |
Outgoing inspection time per roll | 8–12 min (lightbox visual) | <2 min (automated scoring) |
Objective uniformity documentation for buyers | None — inspector sign-off only | Spatial uniformity map and score per roll, archived |
How does NorrStudio measure patchiness objectively when it is inherently a subjective visual assessment?
NorrStudio analyses the statistical distribution of colour intensity values within defined spatial measurement windows across the full fabric width. For each window, it calculates the variance of colour values relative to the window mean a high variance indicates that the window contains both significantly darker and significantly lighter zones, which is the visual perception underlying patchiness. By comparing this spatial variance to the approved fabric standard's baseline variance, the system produces an objective uniformity score that correlates with human visual assessment but is free from inspector subjectivity and fatigue.
Can NorrStudio distinguish between different types of patchiness crease marks versus cloud patches versus preparation faults?
Yes. Each patchiness type has a characteristic spatial morphology. Crease mark patches have a linear, directional structure corresponding to the fold geometry. Cloud patches from exhaust dyeing have a diffuse, irregular gradient with no defined edge. Preparation faults produce fixed-position patches that recur at the same fabric zone across multiple rolls from the same lot. NorrStudio's spatial morphology classifier distinguishes these patterns and reports the likely cause, enabling targeted process intervention rather than generic dye house investigation.
Can the uniformity score be used as objective evidence in buyer-mill disputes over borderline patchiness?
Yes. NorrStudio's spatial uniformity map and roll-level uniformity score provide a quantified, reproducible measurement of patchiness severity that both parties can reference independently of visual assessment. Where buyers and mills agree in advance on the uniformity score threshold corresponding to the pass/fail boundary, the NorrStudio report becomes the binding QC document eliminating the subjective dispute entirely.
At what stage of dye processing should NorrStudio patchiness inspection be deployed?
The optimal deployment point is after washing off and drying when the final dye depth and uniformity of the fabric are established and the patchiness pattern fully visible. For pad-batch processes, inspection at the stenter exit after washing off captures the definitive colour state. For exhaust dyeing, inspection after hydro-extraction and drying on the stenter provides the final assessment before the roll is batched for outgoing QC.
Can NorrStudio identify the specific dye process variable responsible for patchiness on a given roll?
Yes, when integrated with the dye house process control system. NorrStudio correlates the patchiness severity score and spatial pattern for each roll with the process parameters logged during its dyeing cycle liquor ratio, agitation rate, temperature profile, chemical dosing sequence. Recurring correlations between specific parameter deviations and elevated patchiness scores are surfaced as root cause signals, enabling the dye technologist to make targeted recipe or process adjustments rather than general troubleshooting.
Similar Topic
