AppearanceInspection System

All-round AI vision inspection for whole units and parts, covering appearance, dimension and error-proofing with high precision and detection rates in complex mixed-model production

Wide-field small-target fine inspection Flexible mixed-model adaptation Defect recognition in complex environments
Core Scenarios

Three core inspection scenarios

Appearance · dimension · error-proofing — one vision system covering the key quality nodes from whole units to parts

01Surface Defect

Fine appearance inspection

Detects scratches, dents, dirt and paint loss on product surfaces, achieving fine recognition in a wide field and reliably catching flaws the eye can barely see

Whole units · parts
Scratches Dents & bumps Dirt Abrasions Paint loss
Wide-field dirt detection
Dirt detection
Scratch detection
02Dimension

Dimension inspection

High-precision measurement of key dimensions — gaps, flushness, angles, length/width, area, holes and relative positions of structural parts — with automatic vision calibration and global 3D coordinate measurement

0.3mm
Inspection accuracy
99.9%+
Measurement accuracy
3D
3D global measurement
Multi-face camera layout · dimension measurement UI
03Error-proofing

Error-proofing inspection

From labels to sub-assemblies, it verifies assembly consistency and accuracy, eliminating wrong parts, missing parts and mismatched information at the source.

Label consistency
Energy label LOGO Nameplate OCR recognition
Sub-assembly consistency
Panel Buttons Door Screws Assembly error-proofing
Energy label / nameplate verification
Part error-proofing
Imaging to Decision

An integrated chain from imaging to decision

Imaging and algorithms are optimized together for stable operation with high detection and low false-alarm rates under high reflectivity, complex textures and mixed-model production

STAGE 01

Optical imaging

Multi-path lighting and high-resolution imaging reliably reproduce fine features

STAGE 02

AI algorithms

Defect recognition and high-precision measurement suited to complex textures and mixed production

STAGE 03

Real-time decision

Real-time OK / NG output with high detection and low false-alarm rates

STAGE 04

Result integration

Inspection data returns to the line and systems to drive closed-loop handling

Imaging side Algorithm side Optimized together to maintain, under high reflectivity, complex textures and mixed production, high detection rate and low false-alarm rate
Core Advantages

Industrial-grade appearance inspection, with core advantages

Appearance, dimension and error-proofing all meet industrial standards in recognition, flexibility, data and usability

01

Multi-technology fusion

2D and 3D vision combined with AI classification, template matching and OCR for precise wide-field small-target recognition

02

Flexible mixed production

Handles complex mixed flows with multi-variant follow-along inspection and automatic program switching

03

Cloud data processing

Results are stored and uploaded to the management platform in real time, supporting traceability, statistics and analysis

04

Easy to use

The software is general-purpose and friendly, combining flexibly with hardware

Modular & Gates

Modular design with multiple gate mechanisms

Modular design lowers the deployment and maintenance barrier, while four gates check in sequence so every defective unit is intercepted and none leaves the line

Modular Design

Lower deployment and maintenance barrier

Fully modular across mechanics, electrics and software — decoupled modules combine as needed to suit different lines and stations, deploying fast, running stably with few faults

Mechanical module Electrical module Software module
Three-way modular design Mechanics, electrics and software decoupled — combine as needed and deploy fast
Basic computer skills are enough; maintenance is simple No algorithm background required — a low barrier lets on-site staff operate and maintain it independently

Four anomaly-handling gates

FOUR GATES
01

Line stop

Real-time tracking blocks the faulty unit precisely, containing the issue at the current station

02

Auto divert to rework

Diverted to the rework loop and returned to the line after passing re-inspection, without breaking main-line cycle time

03

Defect label

Inspection information is printed and applied automatically, so the anomaly stays visible and traceable with the unit

04

Information gate

Data is written to MES and verified by scanning at line-off, preventing defective units from leaving

Get Started

Bring AI vision to your line — together with HIYAN

WeChat WhatsApp