Steelmaking
Steel wire rope inspection
180
m/min
14–33
mm diameter
95%
accuracy
Broken wires, local wear and strand deformation on a moving rope — no stops for visual checks.
Machine visionQuality control automation
Automated product
quality control
in manufacturing
We catch defects in the flow — before shipment
A computer vision quality inspection system: defect detection, surface flaw identification, photo capture and integration with PCS, MES, ERP and 1C. Designed for your production line.
The problem
Why manual visual inspection
is not enough
Eyeball inspection depends on a human — and that is where defects slip through. Automation removes these bottlenecks.
Human factor
Results depend on the operator, end-of-shift fatigue and attention.
Line speed
At high speed it is impossible to check 100% of items by eye — some defects pass through.
No uniform criteria
Pass / reject judgments differ between shifts, disputed cases go unrecorded.
No statistics
Without photo capture and analytics you cannot trace the root cause of defects or prove quality to the customer.
Defects the system detects
9 defect classes
in a single flow
Optical surface defect inspection and in-line defect detection with photo capture of every decision. Classes are selected and retrained for your product.
Chips
edges, surface
Cracks
breaks, delamination
Coating gaps
paint runs and misses
Deformation
curvature, warping
Contamination
inclusions, buildup
Marking
code, readability
Completeness
assembly, presence
Geometry
dimensions, deviations
Reference match
sample comparison
How it works
From camera to PCS signal —
in a split second
At the core are trained AI models: cameras capture the item, local AI analyzes every frame and makes a pass / reject decision in real time.
- 1
In-line capture
Cameras shoot the item under stable lighting.
- 2
Computer vision model
Finds defect features in the frame.
- 3
Classification
Pass / reject + defect type.
- 4
Interface
Photo, status, event log.
- 5
Integration
Signal to our platform, PCS, MES, ERP, 1C and more.
Results on the line
Quality control becomes
measurable and transparent
−30%
fewer defects thanks to early detection
×5
faster than manual inspection
100%
photo capture of every decision
36 mo
archive and statistics retention
Figures are based on delivered projects; the actual effect depends on product type, initial defect rate and line conditions.
The difference
Not just a camera — a full-featured
machine vision system
A regular camera
- Only shows the picture
- Needs an operator at the monitor
- No defect detection or classification
- No statistics or system integration
I-SOL CV system
- Finds defects and deviations on its own
- Classifies: pass / reject / review
- Photo capture, shift and batch analytics
- Integrates with PCS, MES, ERP, SCADA, 1C
Our experience
Already on the shop floor
We deploy machine vision quality control at operating plants — from steelmaking and rolled metal to agro and welding lines. Every project starts with a line survey and ends with a system running in the flow 24/7 and reporting statuses to enterprise systems. Below are real scenarios from our practice.
Rolled metal
Finished rolled products inspection
7
profiles
OK/NG
QC status
Rebar, rounds, angles, channels, I-beams.
Defect log for the QC department.
Ore processing
Conveyor video analytics
24/7
real time
SCADA
integration
Granulometry, ore contamination and ore color in the flow.
Machine building
Weld seam inspection
5
defect types
up to 95%
accuracy
Pores, undercuts, lack of fusion, overlaps and spatter along the seam. Photo capture of every joint for the QC log.
Agro
Grain sorting and inspection
12
t/h flow
4
impurity classes
Weed and grain impurities, broken and damaged kernels at the elevator — batch quality assessment in the flow.
Trusted by mining and industrial enterprises
Application
Where in-line quality control
is used
Typical scenarios where we deploy automated quality control:
- Weld and joint quality inspection
- Casting quality inspection
- Paint quality inspection
- Rolled metal inspection
- Pipe and rebar quality inspection
- Packaging quality inspection
- Assembly quality inspection
- Machined parts inspection
- In-line production quality control
- Incoming parts inspection
Deployment
8 steps to industrial operation
- 01
Survey
task and line - 02
Equipment
cameras, lighting, server - 03
Data collection
capture and labeling - 04
Model training
CV development - 05
Logic
rules and scenarios - 06
Integration
enterprise systems - 07
Pilot
a real line section - 08
Rollout
launch and support
FAQ
Frequently asked questions
Not quite. We do not sell a boxed product — every project is tailored to the specific task, product type and the client's inspection scenarios. These define the camera setup, lighting, trainable defect classes and rejection rules.
Cameras capture the item in the flow, a trained CV model analyzes the image, classifies the result as pass or reject with the defect type, saves the photo and sends the event to the operator interface and to PCS, MES, ERP or 1C.
Chips and spalling, cracks, coating gaps and paint defects, deformation and geometry deviations, contamination and foreign inclusions, marking errors, incomplete assembly. The exact class set is configured for your product.
On capture conditions (lighting, angle, positioning stability), the volume and quality of training data and correctly defined rules. Our projects reach 95%+ classification accuracy with adaptive retraining on real data.
Yes. The system sends events and statuses to PCS, MES, ERP, SCADA and 1C, supports signal control and line stops on defects, and keeps an event log and photo archive.
No. Inspection runs in the flow without slowing production down. Deployment starts with a survey and a pilot section, and the transition to industrial operation is staged.
No, the system does not override GOST, spec or QMS requirements. It automates visual and optical inspection against defined pass / reject criteria, keeps photo records and an event log — building an evidence base for the QC department.
Manual visual inspection depends on the operator, line speed, lighting and fatigue, and leaves no statistics. A CV system applies uniform criteria to every item, works without gaps in the flow, keeps photo records and shift and batch analytics — complementing, not replacing, instrumental inspection.
Describe your product — we'll assess it in one call
We'll review your line and defects and propose a pilot scheme.
Automating product quality control in manufacturing
I-SOL develops and deploys a machine vision automated quality control system. It is not a boxed product: we select cameras and lighting and train the CV model for your specific line, product type and inspection scenarios. The system automates quality control where manual visual inspection depends on the operator, line speed and end-of-shift fatigue.
Computer vision provides in-line defect detection, optical surface inspection and pass / reject classification with defect typing. Every decision is backed by a photo, and events and statuses are sent to PCS, MES, ERP, SCADA and 1C.
I-SOL visual inspection systems are used in steelmaking and rolled metal production, for weld and joint inspection, casting, painting, pipes and rebar, packaging and assembly, as well as for incoming inspection and in-line quality control. The defect class set is retrained on the plant's real data.
Automated inspection complements instrumental quality control and does not override GOST, spec or QMS requirements: it helps the QC department keep evidence-based records against defined quality criteria. Describe your product and line — we'll propose a pilot scheme and estimate the effect.