基于MobileNetV3编码与U-Net多尺度解码融合的水稻磷素营养诊断OA

Accurate assessment of phosphorus nutrition is essential for optimizing rice growth and im-proving fertilizer management.This study proposes a lightweight image classification model,Mobile-NetV3_U-Net,that integrates a MobileNetV3_large encoder with an improved U-Net decoder for efficient diagnosis of rice leaf phosphorus status.The encoder leverages pretrained weights to accelerate convergence and extracts multi-scale semantic representations,while the decoder adopts an asymmetric multi-scale upsampling strategy with skip connections to progressively fuse shallow and deep features.Unlike the con-ventional symmetric U-Net,the redesigned decoder adjusts channel width and spatial resolution in a com-pact manner,enhancing fine-grained feature characterization and sensitivity to phosphorus deficiency symp-toms.The classification is completed by adaptive average pooling and a fully connected layer.On test sets from the rice tillering and jointing stages,MobileNetV3_U-Net achieved accuracies of 93.33%and 87.40%,respectively,surpassing representative lightweight models including MobileNetV3_large,Ghost-Net,ShuffleNetV2,and EfficientNet_b1.Furthermore,experiments on the Plant Village dataset using grape leaf images confirmed its cross-domain generalization capability.MobileNetV3_U-Net demonstrates a favorable balance between lightweight design and diagnostic accuracy.By enhancing multi-scale feature fu-sion through a tailored decoder structure,it provides an efficient,robust,and transferable framework for in-telligent diagnosis of rice phosphorus nutrition.