A note on “most popular”
We do not have sales or installation statistics for HIKROBOT products, and we will not guess them. This page therefore shows two things we can stand behind: the configurations that most packaging and parts inspection stations end up as, from our engineering work, and the scenarios HIKROBOT itself highlights in its product catalogue (HIKROBOT catalogue 2025 Q4). HIKROBOT product names are used as HIKROBOT publishes them; this website does not claim any official HIKROBOT partnership or distributor status.
Six typical configurations
Each configuration is built on one HIKROBOT device class: a vision sensor, a smart camera, a code reader, or a HIKROBOT camera (area scan or line scan) with a vision controller or PC. A HIKROBOT machine vision camera is connected and its parameters set in HIKROBOT MVS, the camera client software (HIKROBOT camera user manual); the inspection itself then runs in VisionMaster. Where we work, in South Africa, the configuration is chosen after a sample test on your products.
A. Single vision sensor
Hardware: An SC1000, SC2000 or SC3000 vision sensor with its built-in light, digital I/O to the PLC, a reject
Typical checks: Cap present, part present and the right way round, a simple count or character check from one view
Limit: One view, fixed tool set; not for many variants
Presence and counting →B. Single smart camera
Hardware: An SC5000X or SC6000 smart camera running VisionMaster, external light where needed, PLC link, reject with confirmation
Typical checks: Label and date code on one face, OCR/OCV, a deep-learning defect check on one region
Limit: One view; processing time sets the line rate it can keep up with
Label inspection →C. Fixed code reader
Hardware: An ID2000, ID3000 or ID5000 code reader, light or lens cap for the material, PLC protocol, reject for no-read and wrong code
Typical checks: Barcodes, DataMatrix and QR codes on units, cartons and parts; content matched to the batch or order
Limit: Reads codes only; no label or print inspection
Carton and pallet verification →D. Multi-camera station
Hardware: Several HIKROBOT GigE area scan cameras with lights, a VC-series vision controller or PC running VisionMaster, operator screen, PLC network link
Typical checks: All-round labels on round bottles, empty bottle base, sidewall and finish, several checks and many products
Limit: More engineering, wiring and space
Empty bottle inspection →E. Line scan web station
Hardware: HIKROBOT line scan camera(s) (CL series), encoder, line light, controller or PC, defect map and marking or line stop
Typical checks: Film, foil, paper, board and fabric; print defects on the press
Limit: Needs a stable web path and encoder
Web inspection →F. 3D measurement station
Hardware: A DP-series 3D laser profile sensor, encoder, controller or PC running 3D tools
Typical checks: Heights, steps, flatness and volume that a 2D camera cannot measure
Limit: Shiny or dark surfaces and steep walls need testing
Dimensional inspection →The configurations side by side
| Aspect | A. Vision sensor | B. Smart camera | C. Code reader | D. Multi-camera | E. Line scan | F. 3D |
|---|---|---|---|---|---|---|
| Views | One | One | One read window | Several | One line across the web | One profile line |
| Tool set | Fixed, basic | Full VisionMaster incl. deep learning | Code reading | Full VisionMaster incl. deep learning | Web tools, defect map | 3D tools |
| Products / recipes | Few | Many | Many | Many | Many | Several |
| Engineering effort | Lowest | Low to medium | Low | High | High | Medium to high |
Scenarios HIKROBOT highlights
HIKROBOT’s 2025 Q4 catalogue illustrates its smart cameras and code readers with cases from these industries (HIKROBOT catalogue 2025 Q4). They show where the vendor sees its products used; they are not a ranking. The last column says where addaScan works.
| Industry | Cases HIKROBOT shows | addaScan | Our page |
|---|---|---|---|
| Packaging | Package labelling inspection at very high speed; number inspection on food packaging; bottle cap presence detection | Yes: our core | Food and beverage |
| Automotive and auto parts | Error-proofing on multi-model wheels, wheel bearings and parts before assembly; door handle screw installation | Yes | Metal components |
| Manufacturing | Incoming material check on vibrating tray feeders | Yes | Presence and counting |
| Robot guidance | Robot-arm guided picking and multi-point checks | Case by case | Technology |
| Medical | Pill box code traceability; test tube information, classification and liquid level | Not a focus | — |
| Lithium battery, photovoltaic, semiconductor | Battery information and colour checks, cell and cover code reading; PV module and silicon rod codes; wafer font recognition | Not a focus | — |
Which configuration fits the checks on this site
Mapped against the inspection applications described on this site, three configurations cover most of them: a single smart camera (B) for labels, date codes and defects on one face; a fixed code reader (C) for barcode and DataMatrix matching on units and cartons; and a multi-camera station (D) where round containers or several checks need more than one view. A single vision sensor (A) covers the simplest presence checks, and line scan (E) and 3D (F) stations serve the more specialised web and height measurements. This is a design mapping, not market data.
How each is chosen is covered in choosing a HIKROBOT smart camera and choosing a HIKROBOT code reader, and how it is commissioned in the commissioning procedure.
Size the architecture before selecting a controller
Separate image acquisition, computation and product tracking. A camera frame-rate specification answers how images can be produced under stated conditions; it does not specify the time needed to locate a label, run OCR, store an image and confirm a physical reject. The controller must handle the required streams together, including bursts after a conveyor gap. Reserve capacity for fault records rather than benchmarking with storage and communication disabled.
Hypothetical bandwidth calculation: assume four monochrome cameras, each producing 2,432 × 2,048 pixels at 20 frames per second, with one byte per pixel and no image compression. Raw image payload is 4 × 2,432 × 2,048 × 20 = 398,458,880 bytes per second, approximately 398 MB/s or 3.19 Gbit/s before network overhead. That cannot traverse a single 1 Gbit/s link unchanged. This is an arithmetic example, not a measured HIKROBOT configuration or a promise that any controller accepts those inputs. Colour formats, bit depth, regions of interest, shared uplinks and camera timing change the calculation.
Choose separate acquisition links or an appropriate higher-bandwidth architecture, then check storage and CPU/GPU load with the actual solution. Cropping a region can reduce transfer only if supported in the chosen acquisition mode. Saving just failed images lowers long-term storage volume but does not remove the need to process every acquired frame. EMVA’s GenICam material describes the software interface; it does not set a guaranteed throughput for the complete station. (EMVA: GenICam)
| Representative catalogue model | Vendor resolution / maximum frame rate | Architecture implication |
|---|---|---|
| MV-SC5020XM | 1600 × 1216 / 100 fps | Embedded processing; benchmark the complete solution |
| MV-SC5050XM | 2432 × 2048 / 60 fps | Larger image; check field of view and processing time |
| MV-SC5120XM | 4096 × 2944 / 30 fps | More spatial samples with a lower listed maximum rate |
| MV-SC6250M | 5120 × 5120 / 20 fps | Square high-resolution image; C-mount optical selection |
These catalogue rows support a discussion of spatial resolution and image rate, not a ranking of models. Select the full ordering suffix and lens against the working distance and defect contrast. A single large image is useful only if all required surfaces are visible; it cannot replace an opposite-side view hidden by the product itself. (HIKROBOT Smart Product Catalog, 2025 Q4)
Commission acquisition, decisions and recovery as separate layers
| Layer | Configuration steps | Failure test and record |
|---|---|---|
| Acquisition | Assign device identities; set trigger and exposure; check illumination timing | Remove one camera and record how the line responds |
| Inspection | Load approved recipe; link each view to its required check | Feed a missing view or wrong recipe acknowledgement |
| Tracking | Associate results with a physical item; define timeouts and reject position | Stop and restart with products between camera and reject |
| Records | Store reason, recipe, image identity and disposition | Fill the permitted storage allocation and confirm the agreed response |
Treat a partial multi-camera result as incomplete until all required views have arrived or the timeout has expired. An image from the next product must never fill the missing view of the previous one. A product sequence number or equivalent PLC tracking method gives the integrator a way to test this explicitly. Reconcile trigger counts, completed decisions and confirmed rejects, investigating any mismatch rather than resetting it out of the report.
For a South African plant, specify who owns electrical changes, guarded access, compressed air, cleaning and backups. Washdown areas need a hygienic installation appropriate to the site’s food premises controls; R638 supplies the local hygiene context, not a camera enclosure rating or automatic approval of a station. (South Africa: food premises and transport hygiene regulations R638, 2018)
Finally, challenge the decisions using independently identified samples. Good-looking saved images alone are not an acceptance report. Where dimensions are involved, add a measurement-system study against reference values; where defect classification is involved, count false accepts and false rejects separately. NIST’s gauge-study and proportion guidance cover these different forms of evidence. (NIST e-Handbook, 2.4 Gauge R&R studies) (NIST e-Handbook, 7.2.4.1 Confidence intervals for a proportion)