DenseNet201Plus: Cost-effective transfer-learning architecture for rapid leaf disease identification with attention mechanisms

Heliyon, Год журнала: 2024, Номер 10(15), С. e35625 - e35625

Опубликована: Авг. 1, 2024

Plant leaf diseases are a significant concern in agriculture due to their detrimental impact on crop productivity and food security. Effective disease management depends the early accurate detection diagnosis of these conditions, facilitating timely intervention mitigation strategies. In this study, we address pressing need for efficient methods detecting by introducing new architecture called DenseNet201Plus. DenseNet201 was modified including superior data augmentation pre-processing techniques, an attention-based transition mechanism, multiple attention modules, dense blocks. These modifications enhance robustness accuracy proposed DenseNet201Plus model diagnosing related plant leaves. The trained using two distinct datasets: Banana Leaf Disease Black Gram Disease. Through extensive experimentation, evaluated performance terms various classification metrics achieved values 0.9012, 0.9716 accuracy, precision, recall, AUC banana dataset, respectively. Similarly, black gram dataset provides 0.9950, 1.0 AUC. Compared other well-known pre-trained convolutional neural network (CNN) architectures, our demonstrates both utilized datasets. Last but not least, combined strength Grad-CAM++ with interpretability localization areas, providing valuable insights agricultural practitioners researchers make informed decisions optimize

Язык: Английский

---

Paddy Leaf Disease Classification Using EfficientNet B4 With Compound Scaling and Swish Activation: A Deep Learning Approach

IEEE Access, Год журнала: 2024, Номер 12, С. 126426 - 126437

Опубликована: Янв. 1, 2024

Язык: Английский

---

A novel twin vision transformer framework for crop disease classification with deformable attention

Biomedical Signal Processing and Control, Год журнала: 2025, Номер 105, С. 107551 - 107551

Опубликована: Фев. 7, 2025

Язык: Английский

---

Recognition of Multi-Symptomatic Rice Leaf Blast in Dual Scenarios by Using Convolutional Neural Networks

Smart Agricultural Technology, Год журнала: 2025, Номер unknown, С. 100867 - 100867

Опубликована: Фев. 1, 2025

Язык: Английский

---

Rice Disease Detection: TLI-YOLO Innovative Approach for Enhanced Detection and Mobile Compatibility

Sensors, Год журнала: 2025, Номер 25(8), С. 2494 - 2494

Опубликована: Апрель 15, 2025

As a key global food reserve, rice disease detection technology plays an important role in promoting production, protecting ecological balance and supporting sustainable agricultural development. However, existing identification techniques face many challenges, such as low training efficiency, insufficient model accuracy, incompatibility with mobile devices, the need for large number of datasets. This study aims to develop that is highly accurate, resource efficient, suitable deployment address limitations technologies. We propose Transfer Layer iRMB-YOLOv8 (TLI-YOLO) model, which modifies some components YOLOv8 network structure based on transfer learning. The innovation this method mainly reflected four components. First, learning used import pretrained weights into TLI-YOLO significantly reduces dataset requirements accelerates convergence. Secondly, it innovatively integrates new small object layer feature fusion layer, enhances ability by combining shallow deep maps so learn features more effectively. Third, first introduce iRMB attention mechanism, effectively Inverted Residual Blocks Transformers, introduces separable convolution maintain spatial integrity features, thus improving efficiency computational resources platforms. Finally, adopted WIoUv3 loss function added dynamic non-monotonic aggregation mechanism standard IoU calculation accurately evaluate penalize difference between predicted actual bounding boxes, robustness generalization model. final test shows achieved 93.1% precision, 88% recall, 95% mAP, 90.48% F1 score custom dataset, only 12.60 GFLOPS computation. Compared YOLOv8n, precision improved 7.8%, recall rate 7.2%, [email protected] 7.6%. In addition, demonstrated real-time capability Android device 30 FPS, meets needs on-site diagnosis. approach provides support monitoring.

Язык: Английский

---

Plantention: A general-purpose, lightweight and attention-based model for multi-crop leaf disease classification

Results in Engineering, Год журнала: 2025, Номер unknown, С. 105075 - 105075

Опубликована: Апрель 1, 2025

Язык: Английский

---

A hybrid deep learning neural network for early plant disease diagnosis using a real-world Wheat–Barley vision dataset: challenges and solutions

International Journal of Data Science and Analytics, Год журнала: 2024, Номер unknown

Опубликована: Июнь 21, 2024

Язык: Английский

---