Inline AI visual inspection built for small manufacturers
Software: $199–$599 per month per station
Pilot deployments begin ahead of general availability (Q3 2027). The optional $49 pilot-slot deposit is fully refundable.
Real-time 3D render · drag to rotate · final design may vary
What it is
A compact camera with its own light that mounts above a production line.
What it does
Who it's for
Small and mid-size manufacturers, moulders, fabricators and packaging lines.
Look inside
Keep scrolling: it opens up in real time to show the parts that do the work.
Overview
QualityCheck AI is a compact 7 cm industrial camera with an integrated LED ring light and edge AI processor that inspects parts inline on a conveyor. It detects defects, checks assembly completeness, verifies labels, barcodes and QR codes, and triggers a reject mechanism or alerts an operator in real time.
The problem
Small and mid-sized manufacturers rely on manual visual inspection that is slow, inconsistent and tiring, while traditional machine-vision systems are expensive and need specialist programming. Defects escape to customers, rework and scrap eat margins, and there is little data to find root causes.
Find scratches, dents, cracks, flash and discolouration on metal, plastic and moulded parts.
Verify every screw, clip, cap, seal and component is present and correctly placed.
Read and validate labels, barcodes, QR codes, lot numbers and expiry dates.
Check connectors, component presence and orientation on small boards.
Confirm fill level, seal integrity and correct packaging for food and pharma lines.
Track defect types by shift, line and batch to drive root-cause fixes.
Each part is imaged as it passes, lit by the integrated LED ring light and triggered by encoder or sensor.
On-device models check for defects, missing parts and incorrect labels in milliseconds.
Pass/fail is signalled over GPIO to a reject mechanism, stack light or PLC.
Operators see a live view with the defect highlighted on the line display or app.
Results and images feed the dashboard; models are improved over time with new samples.
In the app
QualityCheck Studio lets line engineers train and deploy inspection models from sample images; the dashboard tracks yield and defects across lines and shifts.
Unit 4,811 diverted to reject bin
0.9% in last hour vs 0.3% baseline
3 units in a row, check printer
Model v2.3 loaded, 140 ms cycle
Runs on-device · Hailo-8L
Locates parts and components and verifies presence and position.
Model: YOLOv8 / RT-DETR
Learns what a good part looks like and flags any deviation — no defect examples needed to start.
Model: PatchCore / EfficientAD unsupervised anomaly detection
Outlines the precise defect area for measurement and operator review.
Model: YOLO11-seg / SAM2-assisted labelling
Reads labels, lot codes, expiry dates, barcodes and QR codes.
Model: PaddleOCR (PP-OCRv4) + ZXing
Classifies defect types for trend analysis.
Model: MobileNetV3 classifier
Targets for production units. Subject to change during development.
| AI processor | Hailo-8L (13 TOPS) or NVIDIA Jetson Orin Nano |
|---|---|
| Camera | Industrial 4K global shutter, C-mount lens options |
| Sensors | Trigger input (photo-eye / encoder) |
| Connectivity | Gigabit Ethernet · Wi-Fi 6 · GPIO (reject, stack light, trigger) · Modbus TCP / OPC UA |
| Power | 12–24 V DC, 10–20 W |
| Enclosure | Anodised aluminium industrial housing with integrated LED ring light |
| Dimensions | 7 × 7 × 6 cm |
| Weight | ~450 g |
| Ingress rating | IP65 |
| Sensor | 4K global shutter |
|---|---|
| Lighting | Integrated LED ring light, strobe-synced |
| Lens | Interchangeable C-mount |
| Trigger | Hardware trigger from photo-eye or encoder |
| Processor | Hailo-8L (13 TOPS) / Orin Nano |
|---|---|
| Inspection time | < 50 ms per part (target) |
| Line speed | Up to ~10 parts/s on small conveyors (target) |
| Training | From ~20 good samples to start |
| Digital I/O | 4 in / 4 out GPIO |
|---|---|
| Industrial protocols | Modbus TCP, OPC UA |
| Network | Gigabit Ethernet, Wi-Fi 6 |
| APIs | REST, MQTT |
| Power | 12–24 V DC, 10–20 W |
|---|---|
| Size | 7 × 7 × 6 cm |
| Rating | IP65 |
| Operating temp. | 0 °C to 50 °C |
Built with
Proven, off-the-shelf components our prototypes are built on. Final production parts may differ.
Raspberry Pi 5
Main computer for our edge boxes
Photo: SimonWaldherr,CC BY 4.0, via Wikimedia Commons · AI overlay added
Raspberry Pi AI HAT+ (Hailo AI accelerator)
Runs the vision models on the device
Photo: SparkFun Electronics,CC BY 2.0, via Wikimedia Commons · AI overlay added
NVIDIA Jetson Orin module
Higher-power option for large sites and robots
Photo: 4300streetcar,CC BY 4.0, via Wikimedia Commons · AI overlay added
High-quality camera with C-mount lens
Sharp, fast capture for inspection and decisions
Photo: Laserlicht,CC BY-SA 4.0, via Wikimedia Commons · AI overlay added
Reference photos of commercially available components. IEMI Labs is not affiliated with or endorsed by their manufacturers. Photo credits
Planned contents
QualityCheck Studio lets line engineers train and deploy inspection models from sample images; the dashboard tracks yield and defects across lines and shifts.
Hardware: $2,499–$4,999 per station
Software: $199–$599 per month per station
Every deployment starts with a $500 feasibility study on your sample parts, credited against your first station.
$2,499 + $199/mo
Label, code and presence checks.
$3,499 + $399/mo
Defect and anomaly detection.
$4,999 + $599/mo
High-speed and multi-check lines.
Indicative launch pricing in USD, excluding taxes and shipping. Final pricing is confirmed before any balance is charged.
Illustrative estimates based on internal modelling and typical deployments; not measured results. Actual outcomes depend on site conditions and usage.
Target milestones. We will email reservation holders and pilot partners as they change.
Q4 2026
Pilot applications open; sample-part feasibility testing begins
Q1 2027
Anomaly-detection pipeline validated on pilot customers' parts
Q2 2027
EVT / DVT builds; IP65, EMC pre-compliance and thermal testing
Q3 2027
Pilot lines go live and first production shipments
Q4 2027
Multi-camera inspection and 3D add-on planned
2027 Pilot Program
Pilot partners get early hardware, direct engineering support and preferential launch pricing. Tell us about your site and cameras.
How the pilot program worksA member of the IEMI Labs team will reply within two business days.
No. Applying is free and carries no obligation. If you want to hold a pilot slot, you can place an optional $49 deposit. It is fully refundable any time before shipment — email hello@iemilabs.com or use the refund link in your confirmation email — and it is credited in full against your purchase if you go ahead.
No. QualityCheck Studio guides you through capturing good and defective samples, trains the model, and deploys it to the camera. Our engineers support every pilot setup.
Yes. Inspection runs on the camera in milliseconds. Images stay on your network unless you choose to sync them to the dashboard for analytics and model improvement.
Yes. Pilot applicants can send sample parts or images for a feasibility test so we can confirm the defects are detectable before installation.
QualityCheck accepts a trigger from a photo-eye or encoder and outputs pass/fail over GPIO, Modbus TCP or OPC UA to your reject mechanism, stack light or PLC.
QualityCheck AI is in development. Specifications shown are target specifications and may change as we complete engineering validation. We are targeting a Q3 2027 ship window and will email reservation holders with every major milestone.
We are planning a 2-year limited hardware warranty covering manufacturing defects, with advance replacement for pilot and business customers. Final warranty terms will be published before shipping begins.
A rugged box the size of a paperback that links to your loading-dock cameras.
From $990 per 2-door kit (waived on a 12-month plan)
PilotA small ceiling-mounted camera that watches shop and warehouse floors.
From $399 per edge box
PilotA tiny cube camera that clips onto a shop shelf or ceiling.
From $299 per camera
PilotQualityCheck AI
From $2,499 · Ships Q3 2027