Core Product

CT Garud — AI Defect Detection & Segmentation

Q: What is CT Garud used for?

An AI-powered defect detection and segmentation engine for manufacturing QC. It uses deep learning (YOLO) to detect surface defects, dimensional anomalies, presence/absence, and print-quality issues in real time.

Q: What camera types does it support?

Any GigE Vision industrial camera — Basler, IDS, Hikvision, FLIR and others. No proprietary hardware lock-in; existing line cameras can be used directly.

Q: How quickly can it be deployed?

A typical deployment is 1–5 days from hardware setup to go-live, including camera/lighting setup, model training on your parts, and PLC integration. Retraining for a new variant takes minutes.

Crimson Tech's primary AI vision engine — a deep-learning platform (YOLO-based) that detects, classifies, and segments 10+ defect types simultaneously in a single camera frame. Built for real-time inspection where zero defect escape is non-negotiable.

<50ms inference10+ defect types / frameAny GigE cameraEdge or CloudPLC/Modbus output

Book a Demo → All AI Machine Vision →

Key Features

Multi-Defect Detection

Detects and classifies 10+ defect types in one inference pass — porosity, cracks, scratches, missing features, dimension errors.

<50ms Per Frame

Real-time edge inference (TensorRT-optimized). No cloud round-trip — a verdict before the part leaves the camera field.

Tuned for Zero-Escape

Tuned for zero-escape tolerance and validated on your parts before go-live; <0.2% false-positive rate keeps the line efficient.

Multi-Camera Ready

Orchestrate 8+ cameras in one pipeline — different positions, angles, resolutions — unified into a single decision.

PLC / Modbus Integration

Direct OK/NG to your PLC via Modbus TCP/RTU or digital I/O. No middleware. Auto-reject on defect.

Retrainable by Your Team

Add defect types, variants, or new acceptance criteria in minutes. No vendor call, no extra cost.

Use Cases

Weld bead porosity and pin-hole detection on automotive components
Thread presence and slot dimension verification on machined parts
Surface scratch, dent, and contamination detection on metal stampings
Sticker, label, and badge position verification on assembled products
Black glue bead inspection on automotive glass panels
PCB solder joint and component orientation check

FAQ

An AI-powered defect detection and segmentation engine for manufacturing QC. It uses deep learning (YOLO) to detect surface defects, dimensional anomalies, presence/absence, and print-quality issues in real time.

Any GigE Vision industrial camera — Basler, IDS, Hikvision, FLIR and others. No proprietary hardware lock-in; existing line cameras can be used directly.

A typical deployment is 1–5 days from hardware setup to go-live, including camera/lighting setup, model training on your parts, and PLC integration. Retraining for a new variant takes minutes.