Automated Foreign Material
Detection System

SWIR Vision Technology deployed on iron ore conveyor belts
to detect and reject wooden contaminants in real time —
preventing furnace damage and protecting steel quality.

98%+

Detection Rate

SWIR+IR

Imaging Technology

8–24H

Downtime Prevented

$5M+

Annual Risk Mitigated

Client

The Problem We Solved

Wooden contaminants in iron ore are a silent but critical operational hazard —
invisible to human eyes at conveyor speeds, and catastrophic if they reach the
crusher.

Iron ore conveyor belt in a steel plant

Wooden Contaminants

Pieces of wood, tree bark and organic debris mix into iron ore during mining and transport — making their way onto conveyor belts undetected by standard RGB cameras.

Process risk icon

Process Risk

Foreign organic materials, if undetected, cause furnace damage, reduce steel quality, and trigger unplanned shutdowns and lost production.

Manual inspection icon

Manual Inspection Fails

Human visual inspection on fast-moving ore conveyors is unreliable, inconsistent, and hazardous for operators — no human can sustain 98%+ accuracy 24/7.

High volume throughput icon

High Volume Throughput

Ore conveyors operate continuously at high speed — requiring an automated, real-time detection and rejection solution with zero throughput impact.

Our Four-Step Approach

The system was designed around a complete detect-flag-reject-report loop,
ensuring every contamination event is caught, removed, and documented.

01

Detect step icon

Detect

Automatically identify wooden logs, bark, and organic contaminants on iron ore conveyor belts using SWIR spectral imaging.

02

Flag step icon

Flag

Generate real-time alerts when foreign material is detected — before it enters the processing plant or primary jaw crusher.

03

Reject step icon

Reject

Trigger conveyor stop and mechanical rejection arm to scoop out the detected contaminant with spatial precision.

04

Report step icon

Report

Log all detection events with timestamp, images, and confidence scores for quality audit and model improvement.

The Cost of Undetected Contamination

Wooden logs and organic debris in iron ore can halt an entire plant for up to 24
hours — with cascading damage from conveyor to crusher internals.

Damage Cascade: Conveyor → Crusher

Stage 1

Conveyor Belt

Wood & bark wrap around rollers and idlers, causing belt misalignment and accelerated wear.

Stage 2

Crusher Feed Chute

Large logs wedge in the primary jaw crusher feed throat, triggering immediate blockage and full plant shutdown.

Stage 3

Crusher Internals

Fibrous pulp clogs discharge, jams toggle plates, damages liners and seizes bearings — requiring full dismantling.

Downtime Impact Per Jam Incident

8-24

HOURS TO RESTART

Components at Risk

Crusher Liners & Mantles

Fibrous debris accelerates wear and shortens liner life.

Toggle Plates

Wood jams create sudden overloads that shear the crusher's primary safety component.

Bearings & Shafts

Compacted wood generates off-axis forces causing premature bearing failure.

Conveyor Belt & Idlers

Woody fibre wraps around idlers, causing belt tracking failure and edge damage.

Crusher downtime cost icon

$3–5M+

Annual crusher downtime costs reported by iron ore mines (Int'l Mining, 2021)

Plant restart time icon

8–24H

Plant restart time after each unplanned crusher jam

Intervention point icon

1 Point

Early-stage conveyor detection is the only cost- effective intervention point

SWIR Imaging System

Short-Wave Infrared (SWIR) spectral imaging exploits the fundamental
difference in IR reflectance between organic materials and iron ore to achieve
highly reliable detection.

Why SWIR Works

Spectral contrast icon

Spectral Contrast

Wood and bark strongly reflect IR wavelengths (850–1150nm) while iron ore absorbs them — creating high brightness contrast that makes detection precise.

Surface detection through dust icon

Surface Detection Through Dust

SWIR penetrates surface dust and debris, ensuring visibility of contaminants even in heavy industrial dust environments.

Targeted IR illumination icon

Targeted IR Illumination

IR bar lights at 850–1150nm illuminate the ore belt uniformly, creating consistent imaging conditions independent of ambient light.

Millisecond AI processing icon

Millisecond AI Processing

SWIR feeds are processed by AI in milliseconds, enabling detection at full conveyor operating speed with zero throughput impact.

Key Insight

Under SWIR illumination (850–1150nm), organic materials like wood and bark appear significantly BRIGHTER than iron ore due to their different IR reflectance properties — making detection highly reliable even in dusty, high-throughput environments.

 
SWIR Camera Specifications
TypeShort-Wave Infrared Area Scan
Wavelength850 – 1150 nm
CoverageHigh-res, wide belt coverage
PositioningGlobal Shutter (motion-free)
SpectrumGigE / Camera Link

 

IR Lighting System
TypeIndustrial IR Bar Lights
Wavelength850 nm – 1150 nm
CoverageFull belt width — multi-unit
PositioningAngled for max contrast
SpectrumNon-visible to human eye

How the System Works

End-to-end detection pipeline: from ore conveyor image capture to physical
rejection trigger — all in milliseconds.

01
IMAGE CAPTURE

SWIR Camera Trigger

SWIR camera triggers on belt movement sensor, capturing high-resolution infrared frames.

02
PRE-PROCESSING

Image Enhancement

Normalize, denoise, ROI extraction and contrast enhancement for AI model input.

03
AI INFERENCE

CNN Detection

Deep learning CNN detects and localizes foreign material with spatial precision.

04
DECISION ENGINE

OK / NOK Verdict

Threshold scoring and confidence check produces a final verdict per frame.

05
OUTPUT & ACTION

Rejection Trigger

Alarm trigger, conveyor stop via PLC, and rejection arm deployment.

Pass, OK result

PASS — OK

No foreign material detected. Conveyor continues at full operating speed with no interruption to throughput or production.

Reject, NOK result

REJECT — NOK

Foreign material detected. Conveyor stops, rejection arm deploys, contaminant scooped to rejection bin, belt resumes.

Accuracy & System Constraints

The system achieves industry-leading detection accuracy while operating
within well-defined physical and operational boundaries.

98%+

Detection Rate

Accuracy depends on:

Detection Constraints

Surface Visibility Only

Contaminants must be visible on the ore surface. Buried material cannot be detected by SWIR imaging.

Spectral Visibility Required

Material must exhibit IR reflectance contrast against iron ore in the SWIR spectrum (850–1150nm).

Irregular Sizes

Precision rejection of very small irregular pieces may be challenging — mechanical solution is the most effective intervention.

Detection Constraints

Conveyor Stop Required

Rejection requires belt stoppage — continuous-flow rejection is not feasible with this mechanical design.

Lens Maintenance Schedule

Continuous dust exposure requires a regular camera lens cleaning and inspection protocol.

Environmental Dust

Heavy airborne dust can reduce image clarity over time, requiring periodic system recalibration.

Solution Summary

The SWIR-based Automated Foreign Material Detection System provides a
reliable, real-time solution for identifying and removing wooden contaminants
from iron ore conveyor belts.

SWIR imaging achieves strong spectral separation between wood/bark and iron ore

AI model delivers 98%+ detection accuracy in live production

Mechanical rejection arm stops the belt and removes the contaminant safely

System live in production at an integrated steel plant

Scroll to Top

Request a Demo