Parcel Classification & Detection
Overview
Mounted above induction stations or conveyor lines at sorting hubs, 3D cameras identify parcel types including small parcels, corrugated cartons, foam boxes and poly mailers, alongside parcel physical status: normal or overlapping. As a core data source, the system delivers reliable foundational data for intelligent logistics scheduling and operations.
Key Challenges
-
Diverse parcel materials cause recognition interference: poly mailers, reflective or creased cartons, reflective tape, complex prints, shipping labels and black packaging -
Requires simultaneous classification of parcel type and physical status (normal/overlapping) -
For high-speed conveyors up to 2.5 m/s, the full capture-to-processing cycle must finish within 400 ms
Key Advantages
-
Dual classification of parcel type and status, with ≥99.5% overall accuracy; result output within 400 ms from capture trigger
-
Highly optimized deep learning algorithms, supporting fast training on limited datasets with consistent, repeatable performance
-
Over 30 3D camera models available, for flexible selection across frame rate, accuracy, resolution and FOV
