RDW-YOLO: A Deep Learning Framework for Scalable Agricultural Pest Monitoring and Control

Insects, Год журнала: 2025, Номер 16(5), С. 545 - 545

Опубликована: Май 21, 2025

Due to target diversity, life-cycle variations, and complex backgrounds, traditional pest detection methods often struggle with accuracy efficiency. This study introduces RDW-YOLO, an improved algorithm based on YOLO11, featuring three key innovations. First, the Reparameterized Dilated Fusion Block (RDFBlock) enhances feature extraction via multi-branch dilated convolutions for fine-grained characteristics. Second, DualPathDown (DPDown) module integrates hybrid pooling convolution better multi-scale adaptability. Third, enhanced Wise-Wasserstein IoU (WWIoU) loss function optimizes matching mechanism improves bounding-box regression. Experiments IP102 dataset show that RDW-YOLO achieves [email protected] of 71.3% [email protected]:0.95 50.0%, surpassing YOLO11 by 3.1% 2.0%, respectively. The model also adopts a lightweight design has computational complexity 5.6 G, ensuring efficient deployment without sacrificing accuracy. These results highlight RDW-YOLO’s potential precise in sustainable agriculture.

Язык: Английский

Food sustainability 4.0: harnessing fourth industrial revolution technologies for sustainable food systems

Discover Food, Год журнала: 2025, Номер 5(1)

Опубликована: Май 31, 2025

Язык: Английский