SEUIC (Dongji) / Industrial vision sensors
SEV400: Industrial vision sensors
SEUIC (Dongji) SEV400 is an industrial vision sensor that combines a camera, light and industrial PC for guided AI inspection workflows. The official English page lists a 6 TOPS NPU with up to 12 TOPS, a 1.3 MP image sensor, counting, presence or absence, positioning, identification and OCR; its healthcare industry reference does not establish medical certification or clinical suitability.
Suggested profiles: guided AI inspection and identification stations in manufacturing, healthcare operations and automotive environments
- Product type
- Industrial vision sensor
- Architecture
- Camera + Light + Industrial PC all-in-one
- AI processing
- 6 TOPS NPU, up to 12 TOPS stated by the manufacturer
- Image sensor
- 1.3 MP
- Detection functions
- Counting, presence or absence detection, positioning, identification and OCR
- AI training reference
- Training completes in fewer than 10 sample images through the guided interface, according to the manufacturer
The exact model, regional configuration and commercial terms require project confirmation.
Browse by device capability
SEV400 / Browse by device capability
UROVO is the primary established product resource, with SEUIC (Dongji) and selected platforms used where the capture method, form factor or deployment requirement needs another option. Published models follow an evidence, regional-configuration and RuggedLayer validation path.
Suggested profiles
- Manufacturing presence, count and position checks
- Automotive identification and OCR stations
- Defined healthcare-operations inspection workflows where regulatory scope is separately verified
Browse by device capability
- Camera, light and industrial PC are presented as one vision-sensor architecture for defined inspection stations.
- The manufacturer lists a 6 TOPS NPU with up to 12 TOPS and a 1.3 MP image sensor for AI-powered detection.
- Counting, presence or absence, positioning, identification and OCR provide a starting point for structured inspection workflows.
- Fewer-than-10-sample training is a manufacturer reference; the target mark, background, lighting and acceptance criteria still require testing.
How a device enters the portfolio
Key specifications
Manufacturer-listed configuration; confirm the exact variant before ordering.
Data capture
| Image sensor | 1.3 MP |
|---|
Other configuration
| Product type | Industrial vision sensor |
|---|---|
| Architecture | Camera + Light + Industrial PC all-in-one |
| AI processing | 6 TOPS NPU, up to 12 TOPS stated by the manufacturer |
| Detection functions | Counting, presence or absence detection, positioning, identification and OCR |
| AI training reference | Training completes in fewer than 10 sample images through the guided interface, according to the manufacturer |
| Operation | Guided interface design for simple operation and rapid deployment across multiple inspection scenarios |
| Target industries | Manufacturing, healthcare, automotive and other industrial environments |
A final shortlist depends on the application, scan or RFID workload, operating environment, country, deployment volume and support expectations.
A clear handoff at every layer
Five service layers for a more predictable device program.
RuggedLayer organizes access to multiple manufacturers into a consistent engagement model. Each layer has a defined input, deliverable and evidence boundary.
Service layers
- Confirm the exact camera, lens, light, trigger, I/O, protocol, software image and industrial-PC configuration before quotation.
- Treat healthcare as an industry reference only; establish any medical-device, clinical, privacy and safety requirements separately.
- Pilot representative samples and failure cases under production distance, motion, lighting, contamination and line-speed conditions.
Deployment accessories
- Scope power, mounting, machine wiring, network connectivity, lighting control and protective station hardware around the sensor.
- Version the approved training samples, model settings, inspection recipe, result format and recovery procedure with the handover.
RuggedLayer Test Lab
Define a useful test, not a vague benchmark.
Tell us the labels, tags, application and operating conditions that matter.
- 01
Measure count, presence, positioning, identification and OCR behavior on representative good and bad samples.
- 02
Verify training repeatability, false positives, false negatives, trigger timing, result transport and controller fault handling.
- 03
Freeze the camera, lens, light, model, training set, firmware, power, wiring, protocol and acceptance record before rollout.
Project intake
SEV400 / Start with enough context to make the next step useful.
Choose the closest project intent and describe only non-confidential workflow, application and deployment requirements.