文献类型：期刊文献

英文题名：SCR-YOLO: a small-object detection model for beach litter with spatial and contextual refinements

作者：Fang, Aolin[1];Zhang, Yudong[2,3];Xiao, Xiuchun[4];He, Xinglong[1];Zhu, Daoheng[4]

机构：[1]Guangdong Ocean Univ, Coll Math & Comp Sci, Zhanjiang, Peoples R China;[2]Southeast Univ, Sch Comp Sci & Engn, Nanjing, Jiangsu, Peoples R China;[3]Univ Granada, Data Sci & Computat Intelligence Inst, Granada, Spain;[4]Guangdong Ocean Univ, Coll Elect & Informat Engn, Zhanjiang 524088, Peoples R China

年份：2026

外文期刊名：JOURNAL OF CONTROL AND DECISION

收录：EI(收录号：20262721059260)、ESCI(收录号：WOS:001814566100001)、Scopus(收录号：2-s2.0-105043858204)、WOS

基金：This work was supported in part by National Natural Science Foundation of China under Grant 62472107, in part by Natural Science Foundation of Guangdong Province, China, under Grant 2023A1515011477, in part by the Demonstration Bases for Joint Training of Postgraduates of Department of Education of Guangdong Province under Grant 202205, in part by the Science and Technology Project of Zhanjiang City, under Grant 2025B01061.

语种：英文

外文关键词：Beach litter; object detection; small object; SPDConv; LSKblock

外文摘要：Detecting small-scale litter in beach environments remains challenging due to weak visual cues, complex backgrounds, and significant scale variations. Existing detectors suffer from information loss during downsampling and insufficient contextual modelling, leading to missed detections of small objects. To address these issues, this paper proposes SCR-YOLO, an improved framework built upon YOLOv8 with targeted spatial and contextual refinements. A Space-to-Depth Convolution (SPDConv) preserves fine-grained spatial information during downsampling, while a Large Selective Kernel Block (LSKblock) dynamically expands the effective receptive field and adaptively fuses multi-scale contextual features through spatially weighted selection, enabling more discriminative representation under complex backgrounds. A dedicated small-object detection head further exploits high-resolution features. Evaluated on a self-constructed 11-category beach litter dataset, SCR-YOLO outperforms YOLOv8 and other state-of-the-art methods, achieving gains of 3.1% in mAP50 and 4.9% in mAP50-95, demonstrating its effectiveness within the evaluated beach scenarios.