Principles, Scenarios and Suitability
A container identifier includes an owner code, equipment category, serial number and check digit. Structural validation filters some OCR errors, but passing a checksum does not prove that the correct container is associated with a vehicle or shipment.
The owner code forms part of the equipment-identification structure; the check digit is calculated from other characters. This checks code consistency, not container contents. Separate reading, format validation and shipment matching. A camera could correctly read a neighbouring container but attach it to the wrong passing vehicle.
Engineering must locate the identifier and capture angle without confusing size codes or other signs. Multiple views add evidence but increase mismatching risk with two containers. Keep raw readings from every angle and a persistent trip identifier, with conflicting status for staff review. Do not edit characters to pass the checksum and call them genuine readings. Convenience must not remove the limitation trail needed for later investigation.
Scroll the diagram horizontally to inspect all details.
Illustrative Design Scenario
Illustrative design scenario: read a container side code and associate it with a lane passage. If side and door readings conflict, staff review the originals. Do not automatically choose the code matching the job order to make results appear accurate.
Suppose one side character is ambiguous while a door provides another format-valid value. First confirm both show the same container. Agreement with the job order is not automatically OCR ground truth. Reviewers need each view's image, time and camera position to distinguish a reading error from images of different containers.
Options are adding a view where codes are obscured or referring conflicts to staff. More views reduce some problems but add duplicate data and matching work. Trial two-container trips and closely following vehicles with authorised data. Do not alter real container codes to create examples or use readings to validate customs documents or contents.
Inputs and Prerequisites
Prepare horizontal and vertical codes, old containers, night scenes and reference codes checked on site. Include trip identifiers and camera positions for each view.
A trip ID links images to transport work. Keep time, lane position and container order on the vehicle, with the person confirming the true code. Test rust, shadows and identifier-like signs, not just clearly printed new markings.
Separate raw code, normalised code, check-digit result and trip-matching status. Preserve unknown characters rather than filling them from an order and hiding uncertainty. An angle without a visible code lacks evidence. Repeated readings from one angle are not two independent confirmations: they share the same shadow or rust-related errors.
Trial and Integration Workflow
- 1Locate codes and retain the character readings from each camera view.
- 2Check BIC format and check digits without silently changing raw values.
- 3Combine results per trip and refer conflicts or incomplete codes to people.
Evaluation and Acceptance Criteria
Measure exact whole-code matches per trip and wrong codes that nevertheless pass checksum validation. Test two containers in one image and duplicate views. Character accuracy and checksum-pass rates are not substitutes for end-to-end results.
Exact match requires the complete code and the correct trip association. A correct code attached to the wrong trip is an end-to-end error. Combine multiple views into one container without inflating sample counts from duplicate images.
Staff should sample passed codes and review conflicts. Corrections must preserve raw values and the editor. Report unreadable rates by daylight, darkness and code condition, without silently removing difficult images from scope.
Scroll the diagram horizontally to inspect all details.
Failure Cases and Limitations
Rust, shadows, open doors and unrelated signs can confuse reading. Check digits detect only some error types. Reading an identifier does not certify current ownership, container condition, contents or customs correctness.
Knowing a code does not establish whether the container moved after observation. Link confirmed transport records through the proper process. Yard-location searches are additional tasks needing tracking evidence, not inference from one gate reading.
Code standards guide structure and checks, not certify business documents or all-condition reading. Additional cameras and container types require specific tests, with trip and personal data limited to necessity.
Practical Checklist
- Are raw codes retained before any normalisation?
- Is the code associated with the correct trip and container?
- Are conflicting readings from different views reviewed by people?