ConductVision Enterprise
ConductVision reads the codes and numbers painted or printed on containers, rail wagons and trailers, and logs each read with its image for yard and terminal teams.
5 detectionsConductVision detections on this photo. Photo: AgainErick, CC BY-SA 4.0, cropped.
Optical Character Recognition reads printed and stenciled text from physical objects captured in video streams or images. The system handles ISO 6346 container codes, warehouse labels, serial numbers on equipment, safety signage, and multilingual text across 40+ languages. Unlike document-focused OCR tools, this system is purpose-built for industrial and field environments where text appears on curved surfaces, under uneven lighting, and at oblique angles. Results are structured and searchable, designed to feed directly into inventory, logistics, and compliance systems.
Outcomes
Clerks walk the length of a train to write down each wagon number before it leaves the yard.
The walk takes time on every train, and one misread number files a wagon under the wrong train.
At the gate, staff compare each container or trailer number with the booking by eye.
A swapped pair of digits lets the wrong unit through or holds the right one at the gate.
Yard staff drive the rows to find a trailer or container by its number.
One search can take a full lap of the yard while a driver waits at the gate.
What the model sees
ConductVision detections on this photo. Photo: AgainErick, CC BY-SA 4.0, cropped.
ConductVision detections on this photo. Photo: John Lucas, CC BY-SA 2.0, cropped.
Endpoints
Each endpoint is defined before the pilot starts, with what is counted, its unit and how the results are broken down.
| Endpoint | How it is counted | Unit | Broken down by |
|---|---|---|---|
| PrimaryComplete code rate | Share of containers, wagons and trailers passing a read point with a complete code read, checked against its check digit where the code has one. | Unit:% | By:Read point, unit type |
| SecondaryHand entries left | Codes staff still type by hand because no complete read was made, counted per shift. | Unit:codes per shift | By:Read point |
| Booking mismatches | Codes read at the gate that differ from the expected list, counted once per arrival. | Unit:mismatches per day | By:Gate, hour |
| Train list time | Minutes from the last wagon passing the camera to a complete train list. | Unit:minutes (median) | By:Track |
Each pilot measures ConductVision on your own footage against a sample checked by hand, before anyone relies on its numbers.
Read accuracy
Share of codes read that match a reviewer's transcription of the same image, character for character.
% · by read point, stenciled or printed
Missed codes
Share of codes visible to a reviewer in sampled frames that ConductVision did not find.
% · by camera, distance band
Wrong reads passed
Reads that pass the check digit but differ from the reviewer's transcription, per 1,000 reads.
per 1,000 reads · by read point
How it works
Connect
Mount cameras at container yards, conveyor belts, loading docks, or facility entrances where text needs to be captured. The system calibrates for distance, angle, and expected text regions automatically.
Analyze
A text-detection model localizes text regions in each frame, then a recognition model decodes characters with language-specific dictionaries. Post-processing validates against expected formats like ISO 6346 container codes.
Act
Extracted text is parsed into structured fields (container number, check digit, owner code) and delivered to downstream systems via API. Misreads are flagged by confidence score for manual review.
What it detects
Text
Each block of printed or stenciled text in the frame, boxed and read
Container code
Owner code, serial number and check digit, as separate fields
Wagon number
The number on each rail wagon, in the order the train passes
Trailer number
Fleet and unit numbers painted on trailers and swap bodies
Features
Use cases
Shipping terminals process thousands of containers daily and need accurate code reads for tracking. Automated ISO 6346 reading can replace manual data entry at gate lanes and crane operations.
logisticsshipping
Warehouses need to verify that the right products are in the right locations. Camera-based label reading at dock doors and aisle intersections can catch mis-shipments before they propagate.
logisticsmanufacturingretail
Production lines print lot numbers, date codes, and serial numbers on products. OCR can verify that printed information matches the production order, flagging discrepancies in real time.
manufacturingpharma
Utility companies maintain thousands of poles, meters, and transformers with stamped serial numbers. Mobile and drone-mounted cameras can read asset IDs during inspections, eliminating manual transcription.
utilitiesenergy
Emergency responders and facility managers need to identify hazardous materials quickly. Automated placard reading can extract UN numbers and hazard classes from truck and container signage.
logisticsgovernmentemergency services
Technical overview
Accuracy and speed are measured on your footage and cameras during the pilot, and reported with the conditions they were measured under.
Industries
Tracks forklifts, pallets, trucks and people from dock door to rack.
Checks gloves, masks and goggles, and counts units on filling lines.
Tracks containers, cranes, trucks and vessels across the terminal.
Related
Finds each vehicle and its plate at gates, lanes and car parks.
Classifies, counts and tracks cars, trucks, buses, motorcycles and bicycles.
Checks hard hats, vests, goggles, gloves and masks against each zone's rules.
Pilot
A pilot takes one prioritized outcome, runs on your cameras, reports the endpoints above and ends with a results review and a decision.
