Enhancing Object Detection in Smart Logistics: Integration ofIFrustum-Pointnets Model with Industrial Internet of Things

Abstract

The Industrial Internet of Things (IIoT) is driving unprecedented transformations in the production sector. However, in the field of intelligent logistics object detection, challenges remain in the accuracy and robustness of object recognition. To address these issues, we have introduced the IFrustum-Pointnets model. Specifically, we optimized the threshold selection strategy, choosing a more suitable threshold to enhance the stability and accuracy of target detection. Furthermore, we improved the parallel attention mechanism and replaced the original loss function with Focal Loss to address class imbalance in the dataset, thereby enhancing the model's performance and robustness. We also proposed a new IoT framework to better refine the intelligent logistics system, consisting of four main components: the Big Data Layer, the Edge Data Layer, the IoT Layer, and the Deployment Layer. The proposed algorithm utilizes the Big Data and Edge Data Layers to collect and process data in real time, thus more effectively deploying the IFrustum-Pointnets model. Experimental results on the KITTI 3D Object Detection Benchmark dataset demonstrate that IFrustum-Pointnets surpasses traditional methods in all evaluation metrics. These results not only showcase the potential applications of IFrustum-Pointnets in the field of intelligent logistics but also provide strong support for the development of IIoT technologies.