System for Verifying Inventory Placement from Spatial Images

The problem

In day-to-day operations, shelf images, location plans and stock records are held in separate records and files, preventing the team from reviewing the same case in a single view.

As a result, the link to source evidence can be lost while preparing a location discrepancy review list, leading to repeated checks and uncertainty about the basis for the decision.

Incomplete automation may overlook an obscured label or a temporary move, while poorly controlled automation risks an incorrect change to the stock record.

What we set out to improve

• Time required to prepare a location discrepancy list for review
• Correct routing of cases involving an obscured label or a temporary move to the appropriate specialist queue
• Recording of recommendations corrected or rejected by the warehouse stocktake supervisor

How the system works

1. 01 Retrieve shelf image, location plan and stock record data from an authorised source
2. 02 Verify the source system, record identifier and recency information
3. 03 Transform the fields into the target process schema
4. 04 Prepare a location discrepancy review list draft
5. 05 Compare the draft with business rules and source records
6. 06 Check for exceptions involving an obscured label or a temporary move
7. 07 Obtain the warehouse stocktake supervisor's approval, correction or rejection
8. 08 Write the input, recommendation, changes and final decision to the audit log

Methods we used

• Source-traceable data extraction from shelf images, location plans and stock records
• Structured output that restricts location discrepancy review fields through business rules
• An exception gate that separates cases involving an obscured label or a temporary move
• A queue that shows the source and recommendation together for the warehouse stocktake supervisor
• Duplicate-processing controls and rollback records to mitigate the risk of an incorrect change to the stock record

Where people stay involved

Data and security

Only the fields required for the task are processed from shelf images, location plans and stock records. Source-system permissions remain in force, and personal or commercially sensitive data is minimised before being provided to the model. Every read, recommendation, human amendment and write to the target system is logged.

Who this suits

Frequently asked questions