Identification of Apple Leaf Diseases by Improved Deep Convolutional Neural Networks With an Attention Mechanism

A method of deep network auto-training based on the MTPI auto-transfer learning and a reinforcement learning algorithm for vegetation detection in a dry thermal valley environment

UAV image acquisition and deep learning techniques have been widely used in field hydrological monitoring to meet the increasing data volume demand and refined quality. However, manual parameter training requires trial-and-error costs (T&E), and existing auto-trainings adapt to simple datasets and network structures, which is low practicality in unstructured environments, e.g., dry thermal valley environment (DTV). Therefore, this research combined a transfer learning (MTPI, maximum transfer potential index method) and an RL (the MTSA reinforcement learning, Multi-Thompson Sampling Algorithm) in dataset auto-augmentation and networks auto-training to reduce human experience and T&E. Firstly, to maximize the iteration speed and minimize the dataset consumption, the best iteration conditions (MTPI conditions) were derived with the improved MTPI method, which shows that subsequent iterations required only 2.30% dataset and 6.31% time cost. Then, the MTSA was improved under MTPI conditions (MTSA-MTPI) to auto-augmented datasets, and the results showed a 16.0% improvement in accuracy (human error) and a 20.9% reduction in standard error (T&E cost). Finally, the MTPI-MTSA was used for four networks auto-training (e.g., FCN, Seg-Net, U-Net, and Seg-Res-Net 50) and showed that the best Seg-Res-Net 50 gained 95.2% WPA (accuracy) and 90.9% WIoU. This study provided an effective auto-training method for complex vegetation information collection, which provides a reference for reducing the manual intervention of deep learning.

Multi-kernel inception aggregation diffusion network for tomato disease detection

Tomato leaf diseases significantly impact the yield and quality of tomatoes during cultivation, the main of which are septoria leaf spot, leaf curl virus, verticillium wilt, and early blight. These diseases necessitate prompt detection and management strategies to mitigate their deleterious effects on crop productivity. Due to the considerable scale variations in diseased tomato leaves, accurate and rapid detection and diagnosis remain challenging. To address the detection of tomato leaf diseases at different scales, we propose a real-time detection model incorporating a Multi-kernel Inception Aggregation Diffusion Network. In this paper, (1) We present a Multi-kernel Inception Aggregation Diffusion Network (MIADN) for the feature processing stage, which facilitates the aggregation and diffusion of multi-scale features across hierarchical levels, benefiting the detection of targets at various scales. (2) We present the Multi-kernel Inception Module (MKIM), designed to extract multi-scale object features using diverse convolution kernels, thereby enhancing the model's feature fusion and representation capabilities. (3) We incorporate the efficient FasterNet network at the feature extraction stage to preserve feature diversity and improve the model's ability to extract complex target features. (4) Extensive comparative and ablation experiments demonstrate that our method achieves the mean average precision (mAP50) of 96.6%, surpassing the baseline model by 4.1% and the advanced YOLOv9s model by 2.0%. This method provides an effective solution for high-quality tomato cultivation.

MnasNet-SimAM: An Improved Deep Learning Model for the Identification of Common Wheat Diseases in Complex Real-Field Environments

Deep learning approaches have been widely applied for agricultural disease detection. However, considerable challenges still exist, such as low recognition accuracy in complex backgrounds and high misjudgment rates for similar diseases. This study aimed to address these challenges through the detection of six prevalent wheat diseases and healthy wheat in images captured in a complex natural context, evaluating the recognition performance of five lightweight convolutional networks. A novel model, named MnasNet-SimAM, was developed by combining transfer learning and an attention mechanism. The results reveal that the five lightweight convolutional neural networks can recognize the six different wheat diseases with an accuracy of more than 90%. The MnasNet-SimAM model attained an accuracy of 95.14%, which is 1.7% better than that of the original model, while only increasing the model's parameter size by 0.01 MB. Additionally, the MnasNet-SimAM model reached an accuracy of 91.20% on the public Wheat Fungi Diseases data set, proving its excellent generalization capacity. These findings reveal that the proposed model can satisfy the requirements for rapid and accurate wheat disease detection.

An image deblurring method using improved U-Net model based on multilayer fusion and attention mechanism

The investigation of image deblurring techniques in dynamic scenes represents a prominent area of research. Recently, deep learning technology has gained extensive traction within the field of image deblurring methodologies. However, such methods often suffer from limited inherent interconnections across various hierarchical levels, resulting in inadequate receptive fields and suboptimal deblurring outcomes. In U-Net, a more adaptable approach is employed, integrating diverse levels of features effectively. Such design not only significantly reduces the number of parameters but also maintains an acceptable accuracy range. Based on such advantages, an improved U-Net model for enhancing the image deblurring effect was proposed in the present study. Firstly, the model structure was designed, incorporating two key components: the MLFF (multilayer feature fusion) module and the DMRFAB (dense multi-receptive field attention block). The aim of these modules is to improve the feature extraction ability. The MLFF module facilitates the integration of feature information across various layers, while the DMRFAB module, enriched with an attention mechanism, extracts crucial and intricate image details, thereby enhancing the overall information extraction process. Finally, in combination with fast Fourier transform, the FRLF (Frequency Reconstruction Loss Function) was proposed. The FRLF obtains the frequency value of the image by reducing the frequency difference. The present experiment results reveal that the proposed method exhibited higher-quality visual effects. Specifically, for the GoPro dataset, the PSNR (peak signal-to-noise ratio) reached 31.53, while the SSIM (structural similarity index) attained a value of 0.948. Additionally, for the Real Blur dataset, the PSNR achieved 31.32, accompanied by an SSIM score of 0.934.

Research of segmentation recognition of small disease spots on apple leaves based on hybrid loss function and CBAM

Identification technology of apple diseases is of great significance in improving production efficiency and quality. This paper has used apple Alternaria blotch and brown spot disease leaves as the research object and proposes a disease spot segmentation and disease identification method based on DFL-UNet+CBAM to address the problems of low recognition accuracy and poor performance of small spot segmentation in apple leaf disease recognition. The goal of this paper is to accurately prevent and control apple diseases, avoid fruit quality degradation and yield reduction, and reduce the resulting economic losses. DFL-UNet+CBAM model has employed a hybrid loss function of Dice Loss and Focal Loss as the loss function and added CBAM attention mechanism to both effective feature layers extracted by the backbone network and the results of the first upsampling, enhancing the model to rescale the inter-feature weighting relationships, enhance the channel features of leaf disease spots and suppressing the channel features of healthy parts of the leaf, and improving the network's ability to extract disease features while also increasing model robustness. In general, after training, the average loss rate of the improved model decreases from 0.063 to 0.008 under the premise of ensuring the accuracy of image segmentation. The smaller the loss value is, the better the model is. In the lesion segmentation and disease identification test, MIoU was 91.07%, MPA was 95.58%, F1 Score was 95.16%, MIoU index increased by 1.96%, predicted disease area and actual disease area overlap increased, MPA increased by 1.06%, predicted category correctness increased, F1 Score increased by 1.14%, the number of correctly identified lesion pixels increased, and the segmentation result was more accurate. Specifically, compared to the original U-Net model, the segmentation of Alternaria blotch disease, the MIoU value increased by 4.41%, the MPA value increased by 4.13%, the Precision increased by 1.49%, the Recall increased by 4.13%, and the F1 Score increased by 2.81%; in the segmentation of brown spots, MIoU values increased by 1.18%, MPA values by 0.6%, Precision by 0.78%, Recall by 0.6%, and F1 Score by 0.69%. The spot diameter of the Alternaria blotch disease is 0.2-0.3cm in the early stage, 0.5-0.6cm in the middle and late stages, and the spot diameter of the brown spot disease is 0.3-3cm. Obviously, brown spot spots are larger than Alternaria blotch spots. The segmentation performance of smaller disease spots has increased more noticeably, according to the quantitative analysis results, proving that the model's capacity to segment smaller disease spots has greatly improved. The findings demonstrate that for the detection of apple leaf diseases, the method suggested in this research has a greater recognition accuracy and better segmentation performance. The model in this paper can obtain more sophisticated semantic information in comparison to the traditional U-Net, further enhance the recognition accuracy and segmentation performance of apple leaf spots, and address the issues of low accuracy and low efficiency of conventional disease recognition methods as well as the challenging convergence of conventional deep convolutional networks.

Apple-Net: A Model Based on Improved YOLOv5 to Detect the Apple Leaf Diseases

Effective identification of apple leaf diseases can reduce pesticide spraying and improve apple fruit yield, which is significant to agriculture. However, the existing apple leaf disease detection models lack consideration of disease diversity and accuracy, which hinders the application of intelligent agriculture in the apple industry. In this paper, we explore an accurate and robust detection model for apple leaf disease called Apple-Net, improving the conventional YOLOv5 network by adding the Feature Enhancement Module (FEM) and Coordinate Attention (CA) methods. The combination of the feature pyramid and pan in YOLOv5 can obtain richer semantic information and enhance the semantic information of low-level feature maps but lacks the output of multi-scale information. Thus, the FEM was adopted to improve the output of multi-scale information, and the CA was used to improve the detection efficiency. The experimental results show that Apple-Net achieves a higher mAP@0.5 (95.9%) and precision (93.1%) than four classic target detection models, thus proving that Apple-Net achieves more competitive results on apple leaf disease identification.

A weight optimization-based transfer learning approach for plant disease detection of New Zealand vegetables

Deep learning (DL) is an effective approach to identifying plant diseases. Among several DL-based techniques, transfer learning (TL) produces significant results in terms of improved accuracy. However, the usefulness of TL has not yet been explored using weights optimized from agricultural datasets. Furthermore, the detection of plant diseases in different organs of various vegetables has not yet been performed using a trained/optimized DL model. Moreover, the presence/detection of multiple diseases in vegetable organs has not yet been investigated. To address these research gaps, a new dataset named NZDLPlantDisease-v2 has been collected for New Zealand vegetables. The dataset includes 28 healthy and defective organs of beans, broccoli, cabbage, cauliflower, kumara, peas, potato, and tomato. This paper presents a transfer learning method that optimizes weights obtained through agricultural datasets for better outcomes in plant disease identification. First, several DL architectures are compared to obtain the best-suited model, and then, data augmentation techniques are applied. The Faster Region-based Convolutional Neural Network (RCNN) Inception ResNet-v2 attained the highest mean average precision (mAP) compared to the other DL models including different versions of Faster RCNN, Single-Shot Multibox Detector (SSD), Region-based Fully Convolutional Networks (RFCN), RetinaNet, and EfficientDet. Next, weight optimization is performed on datasets including PlantVillage, NZDLPlantDisease-v1, and DeepWeeds using image resizers, interpolators, initializers, batch normalization, and DL optimizers. Updated/optimized weights are then used to retrain the Faster RCNN Inception ResNet-v2 model on the proposed dataset. Finally, the results are compared with the model trained/optimized using a large dataset, such as Common Objects in Context (COCO). The final mAP improves by 9.25% and is found to be 91.33%. Moreover, the robustness of the methodology is demonstrated by testing the final model on an external dataset and using the stratified k-fold cross-validation method.

MGA-YOLO: A lightweight one-stage network for apple leaf disease detection

Apple leaf diseases seriously damage the yield and quality of apples. Current apple leaf disease diagnosis methods primarily rely on human visual inspection, which often results in low efficiency and insufficient accuracy. Many computer vision algorithms have been proposed to diagnose apple leaf diseases, but most of them are designed to run on high-performance GPUs. This potentially limits their application in the field, in which mobile devices are expected to be used to perform computer vision-based disease diagnosis on the spot. In this paper, we propose a lightweight one-stage network, called the Mobile Ghost Attention YOLO network (MGA-YOLO), which enables real-time diagnosis of apple leaf diseases on mobile devices. We also built a dataset, called the Apple Leaf Disease Object Detection dataset (ALDOD), that contains 8,838 images of healthy and infected apple leaves with complex backgrounds, collected from existing public datasets. In our proposed model, we replaced the ordinary convolution with the Ghost module to significantly reduce the number of parameters and floating point operations (FLOPs) due to cheap operations of the Ghost module. We then constructed the Mobile Inverted Residual Bottleneck Convolution and integrated the Convolutional Block Attention Module (CBAM) into the YOLO network to improve its performance on feature extraction. Finally, an extra prediction head was added to detect extra large objects. We tested our method on the ALDOD testing set. Results showed that our method outperformed other state-of-the-art methods with the highest mAP of 89.3%, the smallest model size of only 10.34 MB and the highest frames per second (FPS) of 84.1 on the GPU server. The proposed model was also tested on a mobile phone, which achieved 12.5 FPS. In addition, by applying image augmentation techniques on the dataset, mAP of our method was further improved to 94.0%. These results suggest that our model can accurately and efficiently detect apple leaf diseases and can be used for real-time detection of apple leaf diseases on mobile devices.

A Performance Comparison of Classification Algorithms for Rose Plants

One of the key roles of Botanists is to be able to recognize flowers. This role has become highly challenging given that the number of discovered flower types are nearing half a million. To support Botanists, Information Technology offers promising solutions. Specifically, machine learning techniques are intrinsically appealing due to being precise enough as required. To this aim, two observations on flower leaves are relevant and leverage flower identification: one, flower plants exhibit unique features in their leaves thus allow distinction of their co-located flowers; two, leaves have a much longer life than flowers thus preserve identity properties longer. This paper proposes the use of machine learning-based identification of rose types by leveraging the features from their leaves. For this purpose, the performance of Naive Bayes, Generalized Linear Model, Multilayer Perceptron, Decision Tree, Random Forest, Gradient Boosted Trees, and Support Vector Machine has been analyzed. This study optimizes the RF model by investigating and tuning its various parameters such as the number of trees, the depth of trees, and splitting criteria. The best results are achieved with gain ratio because it takes more distinct values to avoid the problems associated with Information Gain. Optimizing the number of trees and the depth of trees of RF yield better accuracy than other models. Extensive experiments are performed to analyze the results of ensemble algorithms by using the voting method for each instance. Results suggest that the performance of ensemble classifiers is superior to that of individual models.

Lightweight-Convolutional Neural Network for Apple Leaf Disease Identification

As a widely consumed fruit worldwide, it is extremely important to prevent and control disease in apple trees. In this research, we designed convolutional neural networks (CNNs) for five diseases that affect apple tree leaves based on the AlexNet model. First, the coarse-grained features of the disease are extracted in the model using dilated convolution, which helps to maintain a large receptive field while reducing the number of parameters. The parallel convolution module is added to extract leaf disease features at multiple scales. Subsequently, the series 3 × 3 convolutions shortcut connection allows the model to deal with additional nonlinearities. Further, the attention mechanism is added to all aggregated output modules to better fit channel features and reduce the impact of a complex background on the model performance. Finally, the two fully connected layers are replaced by global pooling to reduce the number of model parameters, to ensure that the features are not lost. The final recognition accuracy of the model is 97.36%, and the size of the model is 5.87 MB. In comparison with five other models, our model design is reasonable and has good robustness; further, the results show that the proposed model is lightweight and can identify apple leaf diseases with high accuracy.