[CMC1] AI-Driven Pallet Identification and Counting for Tile and Roofing Materials

Currently, inventory checks and inbound–outbound reconciliation of tiles and roofing materials at CMC Tiles’ warehouses rely heavily on manual counting by staff and paper-based warehouse documents. This approach is time-consuming, error-prone, difficult to monitor in real time, and highly dependent on individual staff experience. In addition, verifying the exact number and type of pallets loaded onto trucks before dispatch is challenging, especially when pallets are stacked, unevenly arranged, or partially occluded, and when warehouse lighting conditions vary throughout the day.

- Reduce time spent on inventory counting and data entry - Minimize errors in inbound and outbound warehouse operations - Lower manpower costs for periodic inventory audits - Enable alerts when discrepancies occur between physical pallets and warehouse documentation at dispatch points

Manual counting by warehouse staff Reconciliation using paper-based inbound and outbound warehouse documents

- Manual counting by warehouse staff - Reconciliation using paper-based inbound and outbound warehouse documents

- CCU team (order and warehouse management) - Warehouse operations staff - Logistics and transportation teams

Input data: - Images or videos of pallets inside warehouses - Images or videos of pallets loaded on trucks before leaving the factory Identification data: - QR codes attached to pallets - Product information including SKU, size, color, body type, quality grade, and batch/lot number Data types: - Unstructured data: images and videos - Structured data: QR codes, warehouse codes, batch/lot information

Devices: - Smartphones or tablets used by warehouse staff - Optional fixed cameras at warehouse gates or truck weighing stations Operational constraints: - Large warehouse spaces with high shelving - Changing lighting conditions (day/night) - Pallets stacked, overlapping, or unevenly arranged System integration: - Storage of results for future integration with the Bravo system Deployment model: - Open to on-premise, cloud, or hybrid architectures (not yet fixed)