复杂仓储场景下的图像识别优化：MARNnet模型及其多模态自适应机制研究

打开文本图片集

Optimization of Image Recognition in Complex Warehousing Scenarios： Research on the MARNnet Model and Its Multi-modal Adaptive Mechanism

HANLujun1，SHI Yongqiao'，WUYuchen1，WU Zhenyong²（1. 545001；2.信息工程大学 ）（1.GuangxiLoooda;olfgeedegUiof Information Science and Technology，Nanjing 210o44， China）

关键词：智能仓储；图像识别；运动去模糊；注意力机制；MARNnet；多模态融合中图分类号：F252；TP391.41 文献标志码：A DOI： 10.13714/j.cnki.1002-3100.2026.04.036

Abstract： Inintellient warehousingsystems，imagerecognitiontechnologyfaces multiplehallengessuchasunevenillumination， cargoocclusionandmotion blurathghspeeds，eadingtodecreasedrecognitionaccuracyandinsuficientealtimeperforance. Focusing ontheoptimizationofimage recognition incomplex warehousingscenarios，thispaper establishesatheoretical modelof imaginginterferencetorevealtheunderlying mechaismsoffeaturedegradation.Amultimodaladaptiverecognitionnetwork named MARNnetisproposed，icintegateslarableburelstimationadaspatiemporalatetiomehansmstceed toendrobustrecognitionindynamicscenarios.Experimentalresultsdemonstratethattheoptimized systemachievesarecognition accuracy of 96.7% and a processing delay of 45ms in complex scenarios， and improves accuracy by 11.3% and 8.8% compared to YOLOv7 and Detectron2， respectively， efectively supporting the intelligent upgrading of warehousing operations.

Key Words：inteligent warehousing;imagerecogition;motiondeblurring;atentionmechanism;MARNnet;multimoalfusion

0引言

图像识别技术作为仓储自动化的关键感知手段，被广泛应用于货物的定位、分类、质量检测。（剩余5154字）