NVIDIA Inception 2026 CUDA-accelerated Vision Transformers · Built GPU-first
⊕ Computer Vision · TensorRT pipeline · Edge-deployed

Real-time fabric
defect detection
at factory edge.

Human inspectors miss 20–40% of fabric defects at production speed. Our TensorRT-optimised Vision Transformer pipeline detects them in milliseconds — on factory-edge NVIDIA hardware. Built for the 3.6M textile mills that humans inspect today.

The textile industry runs on human eyes that miss defects.

At production speed — 30–120 metres of fabric per minute — even trained inspectors miss between one and four out of every ten defects. The cost compounds downstream as garment rejections, brand returns, and second-grade resale.

20–40%
Defect miss rate by human inspectors at production-line speed
3.6M
Textile mills globally — addressable market for automated CV inspection
$24B
Computer vision in manufacturing market, 2026 (Grand View Research)
10–14%
Typical second-grade rate driven by undetected defects in greige goods

Five stages. Sub-100ms target. Factory-edge.

Cameras over the loom feed a TensorRT-compiled Vision Transformer running on a single NVIDIA edge GPU. Detections are flagged on-line with bounding boxes and defect class — no cloud round-trip, no production-line stop.

01
Capture
Line-scan cameras over the loom. 4K frames at 30 fps. CUDA image preprocessing.
02
ViT inference
TensorRT-compiled Vision Transformer. Defect class + bounding box + confidence.
03
Edge decision
Local rule engine: flag, stop the loom, or mark the bale. No cloud round-trip.
04
Triton serving
NVIDIA Triton Inference Server — concurrent multi-camera, multi-model serving.
05
Audit trail
Per-bale defect history, QA dashboards, exception reports for the mill manager.

Built on the NVIDIA AI stack end-to-end.

No retrofitting. From training on cuDNN-accelerated Vision Transformers to edge-deployed TensorRT inference under Triton, every layer is NVIDIA-native.

PyTorch + Vision TransformersTraining
cuDNNTrain accel
CUDAImage preprocessing
TensorRTEdge optimization
NVIDIA TritonMulti-model serving
NVIDIA Jetson / Jetson OrinFactory edge
NVIDIA NIMEnterprise on-prem
DGX Cloud (Innovation Lab)Fine-tune sprints

Where FabricScan fits in the $1.5T global textile supply chain.

The textile value chain — fibre, yarn, fabric, garment, retail — flows through spinning mills, weaving units, dye houses, and garment factories. We start at the fabric stage, where defects cost the most and humans miss the most.

3.6M
Textile mills globally (ITMF + Wazir Advisors, 2025)
$24B
Computer vision in manufacturing TAM, 2026
$1.5T
Global textile supply chain — fibre to retail
12+
Founder-network source markets across Asia, Europe, MENA, and Latin America

Domain-trained, factory-edge, SME-priced.

Capability FabricScan AWS Rekognition Google Vision
Textile-domain training dataGeneric objectsGeneric objects
Sub-100ms factory-edge inference✓ TensorRTCloud round-tripCloud round-trip
No-cloud / on-prem deployment✓ NIMCloud onlyCloud only
SME-mill pricingPer-image enterprisePer-image enterprise
Per-customer fine-tuning✓ NeMoLimited
Founder-led textile distribution✓ 10yr network

Pilot enquiries open. Q2 2026 cohort.

We are taking on a limited number of textile-mill pilot partners for the Q2 2026 FabricScan TensorRT deployment. If you operate a textile mill anywhere in the world, or you sell to the textile sector — email the founder directly.