Work

Lightweight Detection Methods for Multi-Scale Targets in Complex Scenarios

Year: 2025

Type: article

Abstract: Detecting multi-scale objects in complex scenes is crucial for real-time applications in building safety, where accurate monitoring of safety equipment under challenging conditions is essential. In this paper, we propose LAP-YOLO (Lightweight Aggregate Perception based on "You Only Look Once"), which integrates a Feature-Aware Aggregation (FAA) module to enhance feature representation and a Lightweight Information Diffusion (LID) detection head to improve small-object detection efficiency with minimal computational overhead. Experimental results demonstrate that LAP-YOLO surpasses other real-time models on the safety equipment dataset, achieving a 39% reduction in parameters and a 57% decrease in FLOPs compared to YOLOv8, while maintaining detection accuracy under the same conditions. This lightweight and effective approach presents a promising solution for enhancing construction site safety.

Source: IEEE Access

DOI: 10.1109/access.2025.3560664

Authors: Anjun Yu, Zhichao Rao, Yonghua Xiong, Jinhua She

Institution: China University of Geosciences

Language: English

Topic: Advanced Neural Network Applications

Subfield: Computer Vision and Pattern Recognition

Field: Computer Science

Domain: Physical Sciences

SDGs: Sustainable cities and communities, Industry, innovation and infrastructure

Open Access status: gold

Estimated APC paid: $2,075