A Transformer-Based Detection Network for Precision Cistanche Pest and Disease Management in Smart Agriculture

Plants, Год журнала: 2025, Номер 14(4), С. 499 - 499

Опубликована: Фев. 7, 2025

This study focuses on pest and disease detection in cistanche, proposing a Transformer-based object network enhanced by bridging attention mechanism loss function, demonstrating outstanding performance complex agricultural scenarios. The dynamically fuses low-level details high-level semantics, significantly improving capabilities for small targets backgrounds. Experimental results show that the method achieves an average accuracy of 0.93, precision 0.95, recall 0.92, mAP@50 mAP@75 scores 0.92 0.90, outperforming traditional self-attention mechanisms CBAM modules. These confirm method's ability to overcome challenges such as unclear features target sizes, providing robust support detection. research contributes smart management sustainable development cistanche cultivation while laying solid foundation future intelligence applications.

Язык: Английский

Empowering Smallholder Farmers with UAV-Based Early Cotton Disease Detection Using AI

Drones, Год журнала: 2025, Номер 9(5), С. 385 - 385

Опубликована: Май 21, 2025

Early detection of cotton diseases is critical for safeguarding crop yield and minimizing agrochemical usage. However, most state-of-the-art systems rely on multispectral or hyperspectral sensors, which are costly inaccessible to smallholder farmers. This paper introduces CottoNet, a lightweight efficient deep learning framework detecting early-stage using only RGB images captured by unmanned aerial vehicles (UAVs). The proposed model integrates an EfficientNetV2-S backbone with Dual-Attention Feature Pyramid Network (DA-FPN) novel Symptom Emphasis Module (ESEM) enhance sensitivity subtle visual cues such as chlorosis, minor lesions, texture irregularities. A custom-labeled dataset was collected from fields in Uzbekistan evaluate the under realistic agricultural conditions. CottoNet achieved mean average precision (mAP@50) 89.7%, F1 score 88.2%, early accuracy (EDA) 91.5%, outperforming existing models while maintaining real-time inference speed embedded devices. results demonstrate that offers scalable, accurate, field-ready solution agriculture resource-limited settings.

Язык: Английский

Integration of Diffusion Transformer and Knowledge Graph for Efficient Cucumber Disease Detection in Agriculture

Plants, Год журнала: 2024, Номер 13(17), С. 2435 - 2435

Опубликована: Авг. 31, 2024

In this study, a deep learning method combining knowledge graph and diffusion Transformer has been proposed for cucumber disease detection. By incorporating the attention mechanism loss function, research aims to enhance model’s ability recognize complex agricultural features address issue of sample imbalance efficiently. Experimental results demonstrate that outperforms existing models in detection tasks. Specifically, achieved precision 93%, recall 89%, an accuracy 92%, mean average (mAP) 91%, with frame rate 57 frames per second (FPS). Additionally, study successfully implemented model lightweighting, enabling effective operation on mobile devices, which supports rapid on-site diagnosis diseases. The not only optimizes performance detection, but also opens new possibilities application field

Язык: Английский