Solutions

Plate detection and reading from live camera feeds under conditions that break most systems — motion blur, headlight glare, rain, darkness, oblique angles and dirty plates. The whole pipeline runs on the edge device, so there is no server room, no bandwidth bill and no dependency on a connection that will eventually drop.

• Detection, reading and tracking in a single pass, so a vehicle is one event rather than forty frames
• Plate formats configurable by region rather than hard-coded to one country
• Per-read confidence plus the full evidence frame, so a disputed read can be audited
• Runs fully offline; events queue locally and reconcile when the link returns
• Outputs over REST, MQTT or dry-contact relay into whatever system already exists
• Deploys on Jetson, RK3588 and Snapdragon-class hardware — no GPU server required

Reading container and ISO type codes off moving boxes at gates, cranes and yard portals. Every read is validated against the ISO 6346 check digit before it leaves the device, so a misread is caught at the camera instead of propagating into the terminal system and costing somebody an afternoon.

• Owner code, serial and check-digit validation to ISO 6346 — a failed check is flagged, never guessed
• Multi-camera portal capture so a code on any face of the box is found
• Truck plate correlated with the container ID into one gate event
• Seal presence and door-direction checks in the same pass as the code read
• Visible damage captured and flagged at the gate, where liability is still decidable
• IMO and hazmat placard detection, including the UN number, for dangerous-goods handling
• Structured events into a TOS or WMS over REST, with the evidence images attached
• Built for gates and rail portals where the box does not stop and the light is whatever the sky is doing

Reading wagon identifiers at rail portals and yard entries, where the train does not stop, the number is painted on a curved and frequently filthy surface, and the answer the operator needs is a rake in running order rather than a bag of unrelated numbers.

• Wagons read and sequenced into a rake in running order, with position retained
• UIC and AAR numbering schemes with check-digit validation
• Built for unmanned portals — full speed, day and night, no operator to re-present a missed read
• Correlated with container recognition where the wagon is carrying boxes, so one event covers both
• Structured events into yard management or railway IT over REST

Inspection on a moving line, where the line rate is a commercial commitment rather than an aspiration. Surface defects, presence and absence, orientation, count and dimensional checks — with a deterministic per-part latency instead of a best effort.

• Deterministic latency per part, so the model keeps up with the line rather than the line waiting for the model
• Trained on your parts and your defects, including the few-shot case where real defect examples are scarce
• Thresholds tuned to your false-accept versus false-reject cost, not to a public benchmark
• Every reject logged with its image, so a disputed part can be reviewed rather than argued about
• Runs on an edge box beside the line — no dependency on the plant network staying up

Face and person recognition for consented access control, attendance and occupancy — running entirely on the camera or a local box. Because nothing is sent anywhere, the privacy question has a simple answer instead of a long one.

• Detection, alignment, embedding and matching all on the device; no image or template leaves it
• Enrolment held locally, storing templates rather than photographs where your policy requires it
• Anonymous person detection, counting and dwell time for the many cases where identity is neither needed nor permitted
• Works offline, with no per-query cloud cost as the site count grows
• Deployed with the retention, consent and access controls that GDPR and comparable regimes expect of biometric data