AI-Driven Snag and Pull Inspection for Textile and Apparel Manufacturing
How NorrStudio by NorrSpect enables real-time, automated detection of holes, micro-tears, and fabric breaks on production lines eliminating costly downstream defects before they reach garment assembly.
93%
Reduction in snag-related garment returns and rework
1.2mm
Minimum snag loop height detectable inline
97.8%
Detection accuracy across jersey, interlock, and velour substrates
Overview
Snags and pulls are surface defects that disproportionately affect fabric perceived quality. A single snagged yarn loop raised above the fabric surface even just a millimetre or two is immediately visible to the end consumer and often triggers a product return or brand complaint. In premium knitwear, hosiery, velour, and jersey production, snags are among the top reasons for finished goods rejection at retail incoming inspection.
What makes snags particularly damaging is their origin: they are typically caused by handling equipment conveyor hooks, roller burrs, transfer pins, or packaging machinery meaning they can appear at any stage of production and affect otherwise flawless fabric. NorrStudio, developed by NorrSpect, uses oblique lighting and surface-texture AI models trained specifically for raised-yarn detection to catch snags and pulls inline, before they travel further through the production chain.
About NorrSpect
NorrSpect is a Swedish AI company headquartered in Umeå, Sweden, specialising in industrial visual inspection for high-precision manufacturing. Its NorrStudio platform has been 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. All detection capabilities are defined and validated during the pilot phase using real production data from the client facility..
Industry challenge: why snags and pulls are the hardest surface defects to catch
Unlike holes or stains which create a contrast change visible under frontal lighting snags are three-dimensional surface distortions. A snagged loop raised 1–2mm above the fabric face may cast no visible shadow under flat inspection lighting and be entirely invisible on a standard camera feed. It only becomes apparent under raking or oblique light that reveals surface topology. Human inspectors examining fabric at speed on a standard inspection frame miss the majority of fine snags, particularly on dark, textured, or pile fabrics where the loop blends into the surface texture.
The downstream cost of a missed snag is high: in premium knitwear, a single snagged garment returned by a retail buyer can trigger a quality audit of the entire batch and a renegotiation of unit pricing.
Single yarn snag
One yarn pulled out of the fabric structure and raised above the surface as a loop, caused by contact with a sharp handling point
Pull distortion
A yarn pulled laterally through the fabric structure, creating a puckered or gathered distortion around the pull origin point
Roller burr snag
Repeated snag marks at fixed width intervals caused by a burr or rough spot on a transport or batching roller
Transfer pin mark
A puncture or raised loop pattern left by fabric transfer pins on knitting or finishing machines, creating a repeating defect row
Pile snag (velour / terry)
A pile loop pulled free from the ground structure on velour or terry fabrics, standing above the pile height and visually disrupting the surface uniformity
Packaging catch snag
A snag introduced at roll packaging or dispatch caused by staples, banding clips, or rough cardboard cores often appearing in clusters at roll ends
Solution: NorrStudio AI surface inspection for snags and pulls
NorrStudio uses oblique and raking light configurations where illumination strikes the fabric surface at a low angle to cast micro-shadows from raised yarn loops, making them clearly visible to the AI model even when they are invisible under flat lighting. Deep learning models trained on each client's specific fabric construction and surface topology classify raised loops, pull distortions, and pile disruptions with high accuracy and minimal false positives against textured or pile fabric backgrounds.
Detects single yarn snag loops as small as 1.2mm in height across the full fabric width at production speed
Identifies pull distortions via surface topology mapping, distinguishing pulled structure from intentional texture variation
Detects roller burr and transfer pin snag patterns and correlates them to specific machine positions for maintenance alerting
Operates on pile fabrics including velour, terry, and fleece substrates where standard frontal inspection cameras produce no snag signal
Logs defect roll coordinates for precise cutting room avoidance and roll grading
Flags repeating snag signatures to identify specific handling equipment causing recurring surface damage
Supports premium and luxury fabric lines where near-zero defect escape tolerance is required by brand buyers
Solution
NorrStudio AI Inspection Snag & Pull Module
Inspection scope
Jersey, interlock, velour, terry, fleece, and fine-gauge knit fabric rolls
Hardware
Line-scan cameras, oblique and raking lighting rigs, motion-sync encoder
Output
Real-time snag alerts, annotated roll maps, equipment health signals, PDF QA reports
Integration
ERP / WMS, cutting room CAD systems, maintenance management dashboard
Deployment time
Pilot phase validated on client fabric construction and surface topology before full deployment
Use case: premium knitwear supplier snag elimination for luxury retail
The problem: A premium knitwear supplier producing fine-gauge merino jersey and interlock fabrics for a Scandinavian luxury fashion brand was experiencing an unacceptable snag escape rate at the brand's incoming inspection approximately 4–6% of rolls were being rejected or downgraded on arrival due to surface snags introduced during batching and roll handling. The brand had issued a quality improvement ultimatum threatening to source elsewhere if escape rates were not reduced within two seasons.
The NorrStudio solution: NorrStudio was installed at the batching frame exit and at the roll packaging station. Oblique lighting rigs were configured for the fine-gauge merino construction. Models were trained to distinguish true snag loops from the natural surface texture of the merino yarn. Repeating snag patterns were correlated to a specific batching roller with surface wear, which was identified and replaced within the first week of deployment.
Results:
Metric | Before NorrStudio | After NorrStudio |
|---|---|---|
Snag escape rate at buyer incoming inspection | 4–6% | <0.3% |
Roll rejections per season from snag defects | 18–24 rolls | 0–2 rolls |
Manual inspection time per roll | 10–15 min (oblique lamp) | <2 min (automated) |
Equipment-linked snag identification | Not possible | Faulty roller identified within 1 week of deployment |
Buyer quality audit triggered | Every season | None in 12 months post-deployment |
Traceable roll-level QA documentation | None | 100% archived with annotated snag images |
How does NorrStudio detect snags that are invisible under standard inspection lighting?
NorrStudio uses oblique and raking light configurations where the light source strikes the fabric at a low angle. This causes raised yarn loops even those just 1–2mm in height to cast micro-shadows that are clearly visible to the AI model, even when they produce no contrast signal under flat frontal lighting used in standard inspection frames.
Can NorrStudio detect snags on pile fabrics like velour and fleece without excessive false positives?
Yes. NorrStudio models are trained specifically on each client's pile fabric construction, learning to distinguish the normal variation in pile height and loop structure from a true snag or pulled pile loop. This fabric-specific training is what enables accurate detection on substrates where generic vision systems produce unacceptably high false positive rates.
Can the system identify which piece of handling equipment is causing snags?
Yes. Repeating snag patterns occurring at fixed width intervals or consistent roll positions are flagged as equipment-linked signatures on the NorrStudio dashboard. These patterns indicate a specific roller burr, transfer pin fault, or packaging component causing recurring surface damage, enabling targeted maintenance rather than general line inspection.
Does snag detection work on dark or black fabrics where surface contrast is minimal?
Yes. Because NorrStudio's snag detection relies on raking light to reveal surface topology rather than colour contrast, it performs equally across light and dark fabric colours. Dark fabrics where standard cameras fail are well within the system's detection range.
At what stage of production should NorrStudio be deployed for snag detection?
NorrStudio is most commonly deployed at batching frame exits immediately after knitting or finishing and at roll packaging stations, where the majority of handling-induced snags occur. For high-value fabrics, additional inspection points can be configured at dye house entry and exit to catch snags introduced during wet processing transport.
Similar Topic
