Smart QC System

An AI vision QC and process error-proofing platform built on SOP recognition — self-configured and quick to replicate

Get started
Missing / wrong parts Step verification Anomaly recognition Multi-point operation Traceability
Smart QC System UI: live inspection view, step verdicts and anomaly alerts
SOP Visual Inspection
Core capabilities

Operations recognized, anomalies judged, results traceable

AI vision continuously tracks operator actions, material state and work steps — catching anomalies and safety risks, and retaining a full process record

Assembly recognition AI vision recognizing assembly actions in real time

Work Error-Proofing

Continuously judges picking, placing, assembly and step execution, alerting the moment an error occurs to keep anomalies out of the next process.

Missed picks/places, missing, wrong, reversed parts
Assembly order, skipped steps, incomplete actions
Material placement and work-area checks
Audible, on-screen or device-linked alerts
AI · On-site safety behavior

PPE wearing

On-duty judgment

Restricted-zone intrusion

DANGER ZONE

Anomaly retention

08:12
Station A01Anomaly
14:31
Station B03Pending
16:08
Station C02Reviewed

On-site safety behavior recognition

Fixed cameras continuously recognize PPE wearing, on-duty status and hazard-zone behavior, alerting in time to any breach of site safety rules.

Helmet, vest, mask and glove detection
On-duty, off-duty and unmanned checks
Intrusion into machine, hazard and no-entry zones
Anomaly records kept for query and review
Cycle logging & anomaly tracing
Step 019.4s
Step 0213s
Step 0311s
Total cycle44s
Per-step execution time
Latest records
Station A01 · Step 03 Normal
Station A02 · Step 02 Normal
Station B01 · Step 04 Handled
Station C03 · Step 01 Normal

Cycle logging & process traceability

Continuously logs the result, duration and anomaly footage of every step, turning production into objective, searchable, replayable records.

Logs step time, wait time and full cycle
Links model, station, step, time and image
Replayable anomalies cut recall and back-and-forth
Grounds for training, process and line improvement
Real-time recognition On-the-spot alerts Process trail
Smart Quality Control Platform
Build and run it yourself

Factories build and run their own AI inspection tasks

Define requirements, orchestrate SOP flows, label samples, train models, deploy to stations and adjust later — all in one platform, so model, process and rule changes land faster

SOP flow & rule config Self-defined inspection: code scan, pick-and-place checks, step verification and result retention

Define inspection rules yourself, to your process

Engineers turn floor requirements into executable inspection tasks around the real process

Inspection targets
Inspection regions
Pass conditions
Alert modes
Multi-model & process config Flow-based inspection orchestration: visual composition of the inspection library and step logic

Build complete SOP inspection flows by visual orchestration

Visually compose scanning, picking, placing, assembly, action recognition, step verification and result logging — turning paper SOPs into system-executable on-site flows

Flow orchestration
Step logic
Device linkage
Template reuse
Defect sample labeling Defect sample labeling: image list, inspection labels and labeling history

Train and optimize models yourself, from on-site samples

Collect on-site samples for new materials, actions and product models, then label data, train models, validate results and iterate — all in the platform

Sample collection
Data labeling
Model training
Result validation
Release & station rollout
Deployed
Final line A01
Existing station
StatusNormal
VersionSOP V2.3
CameraOnline
PLCOnline
Replicated
Assembly line B03
Similar-station replication
Source planFinal line A01
VersionSOP V2.3
Config reuse92%
StatusNormal
Released
New line C02
New line
FlowSynced
RulesSynced
ModelPublished
DeviceOnline

Deploy validated inspection plans to on-site stations

Publish validated flows and rules to designated stations; proven inspection plans can be reused on similar stations and new lines

Station release
Device onboarding
Version control
Cross-line replication
Fast to changeAdjust and validate new models in the platform
Less reworkExisting flows, rules, models and parameters reuse across similar scenes
Less dependencyDaily config, training and deployment are done by plant engineers themselves
Build in-house capabilityOn-site samples, rules, model versions and process data keep accumulating
Key application scenarios

Error-proofing and QC across the whole production process

From changeover and material picking to anomaly response and traceability — tasks configured around your process

Model check→Prep→Step check→State check→Response→Traceability

Product & model check

01 · Changeover

Identify product, model and batch, then auto-load the matching SOP flow and inspection tasks.

BarcodeModel matchTask load

Material, tool & fixture check

02 · Prep

Verify picked materials, tools and fixtures match the current product and station.

Wrong partMissing partWrong toolWrong fixture

Step & sequence check

03 · Step check

Continuously track picking, placing, assembly and fastening to catch missed, out-of-order or incomplete steps.

Missed stepWrong orderIncomplete

Assembly state check

04 · Result

Check part count, orientation, position and the finished state of the assembly.

MissingReversedMisalignedShort count

Alerts & line response

05 · Response

Flag anomalies instantly and trigger alarms, blocking, line stop or release as configured.

PromptAlarmBlockPLC link

Final check & traceability

06 · Records

Re-check key items and retain results, anomaly images, step logs and timestamps.

Final checkAnomaly searchQuality trace
Cross-industry fit

For lines with manual assembly, operator error-proofing and process quality needs

Auto partsPick · Fasten · Assemble Wire harnessTerminal · Connect · Order AppliancesParts · Label · Recheck 3CSmall parts · Screws · Connectors New energy · Packaging · More
Solution Comparison
Comparison

Find the build approach that fits your stage

Pick the inspection solution that best fits your stage, by station scale, model change and self-build needs

Dimension
Standard rollout
Smart QC System
Goal
Fast error-proofing at key stations
In-house capability to build and run inspection
Scope
One or a few key stations
Multi-model, multi-point, multi-line
Build method
Delivered by HIYAN to site needs
Built in-platform by plant engineers
Inspection definition
Rules set per project needs
Self-define targets, regions, steps, conditions
Model capability
Trained by the project team
Self label, train, validate, update
Deployment
Deployed per station
Self-publish and reuse flows, rules, models
Long-term value
Solves the current station
Less rework; builds in-house capability
FAQ
Answers

FAQ

Answering common questions on recognition, on-site deployment, system integration and ongoing maintenance

To your process, it can detect missed picks, missed places, missing, wrong and reversed parts, wrong assembly order, skipped steps, incomplete actions and material-placement offsets. It also extends to PPE-wearing detection (helmets, vests) and safety-behavior recognition such as off-duty and restricted-zone intrusion. The exact scope is configured to site requirements.
Work mainly involves adding cameras, lighting, industrial controllers and on-site signal integration; scope depends on station space, cycle time, lighting and existing equipment. The timeline is assessed from inspection content, station count and site conditions. HIYAN completes a site survey and plan sign-off before implementation to minimize disruption to normal production.
As required, it connects industrial cameras, general cameras, lighting, PLCs, scanners, sensors, control cards, audible-visual alarms and line signals. It also links with MES and other systems for model calling, task dispatch, result upload and process traceability. Exact interfaces are confirmed against on-site devices and protocols.
Engineers need a basic grounding in automation, equipment, process or computer operation, but not full AI-algorithm R&D. The platform is visual and configuration-based, with clear procedures for inspection definition, flow orchestration, data labeling, model training and station release. HIYAN provides training and project coaching; once engineers grasp the basics they get up to speed quickly and gradually handle daily config, training and tuning.
Engineers configure inspection targets, work steps, regions, conditions and models per model in the platform. Similar models reuse existing flows, rules and models, adjusting and validating from the prior config. For very different products or complex scenes, HIYAN continues to support sample preparation, model optimization and on-site tuning.
Lighting changes, occlusion, camera angle, action variation and cycle time all affect recognition, so the site is assessed before implementation. HIYAN optimizes camera position, lighting, inspection regions and on-site samples, and improves stability through multi-scene training and real-production validation. New situations later can be handled by adding samples and continuously refining the model.
After go-live, plant engineers handle common inspection config, model changes, data labeling, model training, station release and run monitoring themselves. HIYAN provides ongoing platform training, complex-scene optimization, upgrades, troubleshooting and technical support, helping build in-house vision capability while keeping the system stable long-term.
For one or a few critical stations with a clear inspection target, a standard rollout delivered by HIYAN gets error-proofing running fast. Once you have multiple models, stations or lines — or want to define, train, deploy and maintain solutions in house — Smart QC System fits better: less repeat development, less vendor dependency, and inspection capability that replicates across stations and lines.

More FAQs: selection, deployment, support →

Get Started

Bring AI vision to your line — together with HIYAN

Contact us 400-6936913
WeChat WhatsApp