Efficient diagnosis of apple diseases and pests is crucial to the healthy development of the apple industry. However, the existing single-source image-based classification methods have limitations due to the constraints of single-source input image information, resulting in low classification accuracy and poor stability. Therefore, a classification method for apple disease and pest areas based on multi-source image fusion is proposed in this paper. Firstly, RGB images and multispectral images are obtained using drones to construct an apple diseases and pests canopy multi-source image dataset. Secondly, a vegetation index selection method based on saliency attention is proposed, which uses a multi-label ReliefF feature selection algorithm to obtain the importance scores of vegetation indices, enabling the automatic selection of vegetation indices. Finally, an apple disease and pest area multi-label classification model named AMMFNet is constructed, which effectively combines the advantages of RGB and multispectral multi-source images, performs data-level fusion of multi-source image data, and combines channel attention mechanisms to exploit the complementary aspects between multi-source data. The experimental results demonstrated that the proposed AMMFNet achieves a significant subset accuracy of 92.92%, a sample accuracy of 85.43%, and an F1 value of 86.21% on the apple disease and pest multi-source image dataset, representing improvements of 8.93% and 10.9% compared to prediction methods using only RGB or multispectral images. The experimental results also proved that the proposed method can provide technical support for the coarse-grained positioning of diseases and pests in apple orchards and has good application potential in the apple planting industry.

Wheat diseases threaten yield and food security, highlighting the need for rapid, accurate diagnosis in precision agriculture. However, current remote sensing methods often lack hierarchical spectral semantic perception or rely on single-source data and simple fusion, limiting diagnostic performance. To address these challenges, this study proposed MSFNet, a novel multi-source fusion network with enhanced hierarchical spectral semantic perception, to achieve the precise regional classification of wheat diseases. Specifically, a multi-source fusion module (MSFM) was developed, employing a dual-branch architecture to simultaneously enhance spatial–spectral semantics and comprehensively explore complementary cross-modal features, thereby enabling the effective integration of critical information from both modalities. Furthermore, a hierarchical spectral semantic fusion module (HSSFM) was developed, which employs a pyramid architecture integrated with attention mechanisms to fuse hierarchical spectral semantics, thereby significantly enhancing the model’s hierarchical feature representation capacity. To support this research, we constructed a new multispectral remote sensing dataset, MSWDD2024, tailored for wheat disease region diagnosis. Experimental evaluations on MSWDD2024 demonstrated that MSFNet achieved 95.4% accuracy, 95.6% precision, and 95.6% recall, surpassing ResNet18 by 6.0%, 6.0%, and 5.8%, respectively, and outperforming RGB-only models by over 12% across all metrics. Moreover, MSFNet consistently exceeded the performance of existing state-of-the-art methods. These results confirm the superior effectiveness of MSFNet in remote sensing-based wheat disease diagnosis, offering a promising solution for robust and accurate monitoring in precision agriculture.

The widespread application of convolutional neural networks (CNNs) has led to significant advancements in object detection. However, challenges remain in achieving efficient and precise extraction of critical features when applying typical CNN-based methods to remote sensing detection tasks: (1) The convolutional kernels sliding horizontally in the backbone are misaligned with the features of arbitrarily oriented objects. Additionally, the detector shares the features extracted from the backbone, but the classification task requires orientation-invariant features while the regression task requires orientation-sensitive features. The inconsistency in feature requirements makes it difficult for the detector to extract the critical features required for each task. (2) The use of deeper convolutional structures can improve the detection accuracy, but it also results in substantial convolutional computations and feature redundancy, leading to inefficient feature extraction. To address this issue, we propose a Task-Sensitive Efficient Feature Extraction Network (TFE-Net). Specifically, we propose a special mixed fast convolution module for constructing an efficient network architecture that employs cheap transform operations to replace some of the convolution operations, generating more features with fewer parameters and computation resources. Next, we introduce the task-sensitive detection module, which first aligns the convolutional features with the targets using adaptive dynamic convolution based on the orientation of the targets. The task-sensitive feature decoupling mechanism is further designed to extract orientation-sensitive features and orientation-invariant features from the aligned features and feed them into the regression and classification branches, respectively, which provide the critical features needed for different tasks, thus improving the detection performance comprehensively. In addition, in order to make the training process more stable, we propose a balanced loss function to balance the gradients generated by different samples. Extensive experiments demonstrate that our proposed TFE-Net can achieve superior performance and obtain an effective balance between detection speed and accuracy on DOTA, UCAS-AOD, and HRSC2016.

Transformers and their variants have demonstrated remarkable performance in image classification tasks. MobileViT, as a lightweight architecture, has gained significant attention for real-time image processing due to its computational efficiency and adaptability to resource-constrained environments. It exhibits strong robustness to variations in shape, scale, and viewpoint. However, it still struggle to capture dimensional features and internal relationships within objects. In this paper, we propose a novel feature extraction module called persistent homology with channel attention (PHCA) for image classification. Persistent homology is employed to extract topological features and capture complex spatial relationships across multiple image scales. A channel attention (CA) mechanism is integrated with PH to direct the model's focus toward informative channels, thereby enhancing relevant features while suppressing noise. Furthermore, experiments on fusing PH-based and original features at different network layers indicate that early fusion yields the best performance. To evaluate the effectiveness of our feature extraction method, we conducted experiments on diverse datasets. The results demonstrate that our approach outperforms baseline models in classification accuracy, highlighting the potential of integrating topological features with deep learning for advanced image classification tasks.