Title: HYolo: Leveraging Hypergraph Learning for Intelligent IoT Object Detection

Abstract

This study introduces HYolo, a novel framework for intelligent object detection within the Internet of Things (IoT) that embeds hypergraph learning capabilities directly into the YOLO architecture. Conventional YOLO models are typically limited in their ability to capture pairwise feature interactions, often struggling to represent the complex, high-order relationships between objects and their contextual features. To overcome these constraints, HYolo utilizes hypergraph learning to extract richer contextual dependencies, thereby enhancing the quality of object representation.

Extensive testing on the COCO dataset reveals that HYolo significantly outperforms standard baseline YOLO models. Specifically, the system delivers an approximate 12% gain in mAP@50, alongside notable enhancements in both overall detection accuracy and robustness. By effectively modeling high-order feature relationships, HYolo offers superior contextual comprehension and more dependable detection outcomes in IoT settings. These findings suggest that incorporating hypergraph learning into object detection workflows represents a highly promising avenue for developing intelligent, context-aware vision systems for IoT applications.

Source: arXiv Generated at: 2026-06-04 00:00:00 UTC