Convolutional neural networks have an immense impact on computer vision tasks. However, the accuracy of convolutional neural networks on a dataset is tremendously affected when images within the dataset highly vary. Test images of plant leaves are usually taken in situ. These images, apart from the region of interest, contain unwanted parts of plants, soil, rocks, and/or human body parts. Segmentation helps isolate the target region and a deep convolutional neural network classifies images precisely. Therefore, we combined edge and morphological based segmentation, background subtraction, and the convolutional neural network to help improve accuracy on image sets with images containing clean and cluttered backgrounds. In the proposed system, segmentation was applied to first extract leaf images in the foreground. Several images contained a leaf of interest interposed between unfavorable foregrounds and backgrounds. Background subtraction was implemented to remove the foreground image followed by segmentation to obtain the region of interest. Finally, the images were classified by a pre-trained classification network. The experimental results on two, four, and eight classes of datasets show that the proposed method achieves 98.7%, 96.7%, and 93.57% accuracy by fine-tuned DenseNet121, InceptionV3, and DenseNet121 models, respectively, on a clean dataset. For two class datasets, the accuracy obtained was about 12% higher for a dataset with images taken in the homogeneous background compared to that of a dataset with testing images with a cluttered background. Results also suggest that image sets with clean backgrounds tend to start training with higher accuracy and converge faster.
Tomato disease classification based on images of leaves has received wide attention recently. As one of the best tomato disease classification methods, the convolutional neural network (CNN) has an immense impact due to its impressive performance. However, better performance is verified by independent identical distribution (IID) samples of tomato disease, which breaks down dramatically on out-of-distribution (OOD) classification tasks. In this paper, we investigated the corruption shifts, which was a vital component of OOD, and proposed a tomato disease classification method to improve the performance of corruption shift generalization. We first adopted discrete cosine transform (DCT) to obtain the low-frequency components. Then, the weight of the feature map was calculated by multiple low-frequency components, in order to reduce the influence of high-frequency variation caused by corrupted perturbation. The proposed method, termed as a multiple low-frequency attention network (MLFAnet), was verified by the benchmarking of ImageNet-C. The accuracy result and generalization performance confirmed the effectiveness of MLFAnet. The satisfactory generalization performance of our proposed classification method provides a reliable tool for the diagnosis of tomato disease.
1 January 2022 in International Journal of Information and Education Technology
Information and Communication Engineering is the forefront of science and technology competition among major countries in the world, which is characterized by rapid updating of knowledge system, extensive cross-cutting of scientific research directions and high requirements on the innovation ability of talents. This paper puts forward the mode of cultivating graduate innovative talents in information and communication engineering subject, which combines morality and talent, science and education. Guided by the spirit of scientists, to carry out ideological and political education in an all-round and multidimensional way, the training mode aims to comprehensively enhance the motivation of graduate students and aims to enhance the ideal and faith of serving and developing a powerful country through science and innovation, with high-level scientific research projects and innovative achievements as the guidance, the integration of science and education and multiple measures to cultivate graduate students' scientific research innovation practical ability. The practice shows that the moral education effect of the talent training mode proposed in this paper is remarkable, which greatly improves the motivation of study and enthusiasm for innovation of postgraduates and strengthens the ideal and faith of a strong country in science and technology. Moreover, the innovation ability of the graduate students has been significantly improved, and the graduate students have made great progress in systematic thinking, creative thinking and practical ability. Besides, they have made more high-level innovation achievements.
Crop disease classification constitutes a significant and longstanding challenge in the domain of agricultural and forestry sciences. Frequently, there is an insufficient number of samples to accurately discern the distribution of real-world instances. Leveraging the full potential of the available data is the genesis of our approach. To address this issue, we propose a supervised image augmentation technique—Negative Contrast. This method employs contrast images of existing disease samples, devoid of disease areas, as negative samples for image augmentation, particularly when the samples are relatively scarce. Numerous experiments demonstrate that the employment of this augmentation method enhances the disease classification performance of several classical models across four crops—rice, wheat, corn, and soybean, with an accuracy improvement reaching up to 30.8%. Furthermore, the comparative analysis of attentional heatmaps reveals that models utilizing negative contrast focus more accurately and intensely on the disease regions of interest, thereby exhibiting superior generalization capabilities in real-world crop disease classification.
Crop disease classification has always been a critical and persistent problem in the field of agricultural and forestry sciences, where often we do not have access to a sufficient number of samples to know the distribution of real-world samples. How to make full use of the existing data is the starting point of our thinking. To address this problem, this paper proposes a supervised image augmentation method Negative Contrast, which uses the contrast images of existing disease samples after removing disease areas as negative samples for image augmentation when samples are relatively scarce. Numerous experiments have shown that several classical models using this augmentation method have improved in disease classification of four crops, rice, wheat, corn, and soybean, with a maximum accuracy improvement of 30.8%. In addition, the comparative analysis of attentional heat map shows that the model using Negative Contrast is more accurate and intense on the area of interest of diseases, and thus reflects better generalization ability in real-world disease classification. Our dataset and codes can be found in https://www.kaggle.com/datasets/w970704112/corn-wheat-rice-soybean and https://github.com/hiter0/contrastaug .
Pest detection in agriculture faces the challenge of adapting to new pest species while preserving the ability to recognize previously learned ones. Traditional model fine-tuning approaches often result in catastrophic forgetting, where the acquisition of new classes significantly impairs the recognition performance of existing ones. Although knowledge distillation has been shown to effectively mitigate catastrophic forgetting, current research predominantly focuses on feature imitation, neglecting the extraction of potentially valuable information from responses. To address this issue, we introduce a response-based distillation method, called adaptive response distillation (ARD). ARD incorporates an adaptive response filtering strategy that dynamically adjusts the weights of classification and regression responses based on the significance of the information. This approach selectively filters and transfers valuable response data, ensuring efficient propagation of category and localization information. Our method effectively reduces catastrophic forgetting during incremental learning, enabling the student detector to maintain memory of old classes while assimilating new pest categories. Experimental evaluations on the large-scale IP102 pest dataset demonstrate that the proposed ARD method consistently outperforms existing state-of-the-art algorithms across various class-incremental learning scenarios, significantly narrowing the performance gap compared to fully trained models.