Parcel Re-identification Method Based on Graph Neural Networks

• 1、School of Modern Post,Beijing University of Post and Telecommunications,Beijing 100876

Abstract：On the parcel sorting line, when parcels containing prohibited items are detected by the security inspection machine, they need to be re-identified and located by a robot from another camera on the conveyor belt. The re-identification process of parcels containing prohibited items is often affected by factors such as occlusion or changes in posture, leading to decreased recognition accuracy. To address this issue, this paper proposes a parcel re-identification method based on graph neural networks. Firstly, attention mechanisms are integrated into the network for extracting more representative global features of parcels. Secondly, based on graph neural networks, parcels are parsed into views, and a graph structure is constructed using the extracted features. Finally, experimental analysis is conducted on the parcel re-identification process. Experimental results demonstrate that the proposed method can re-identify target parcels containing prohibited items across camera scenes.

Keywords： Deep learning Parcel reidentification Feature extraction