Accurate classification of plant disease by farming robot cameras can increase crop yield and reduce unnecessary agricultural chemicals, which is a fundamental task in the field of sustainable and precision agriculture. However, until now, disease classification has mostly been performed by manual methods, such as visual inspection, which are labor-intensive and often lead to misclassification of disease types. Therefore, previous studies have proposed disease classification methods based on machine learning or deep learning techniques; however, most did not consider real-world plant images with complex backgrounds and incurred high computational costs. To address these issues, this study proposes a computationally effective residual convolutional attention network (RCA-Net) for the disease classification of plants in field images with complex backgrounds. RCA-Net leverages attention mechanisms and multiscale feature extraction strategies to enhance salient features while reducing background noises. In addition, we introduce fractal dimension estimation to analyze the complexity and irregularity of class activation maps for both healthy plants and their diseases, confirming that our model can extract important features for the correct classification of plant disease. The experiments utilized two publicly available datasets: the sugarcane leaf disease and potato leaf disease datasets. Furthermore, to improve the capability of our proposed system, we performed fractal dimension estimation to evaluate the structural complexity of healthy and diseased leaf patterns. The experimental results show that RCA-Net outperforms state-of-the-art methods with an accuracy of 93.81% on the first dataset and 78.14% on the second dataset. Furthermore, we confirm that our method can be operated on an embedded system for farming robots or mobile devices at fast processing speed (78.7 frames per second).

The segmentation of crops and weeds from camera-captured images is a demanding research area for advancing agricultural and smart farming systems. Previously, the segmentation of crops and weeds was conducted within a homogeneous data environment where training and testing data were from the same database. However, in the real-world application of advancing agricultural and smart farming systems, it is often the case of a heterogeneous data environment where a system trained with one database should be used for testing with a different database without additional training. This study pioneers the use of heterogeneous data for crop and weed segmentation, addressing the issue of degraded accuracy. Through adjusting the mean and standard deviation, we minimize the variability in pixel value and contrast, enhancing segmentation robustness. Unlike previous methods relying on extensive training data, our approach achieves real-world applicability with just one training sample for deep learning-based semantic segmentation. Moreover, we seamlessly integrated a method for estimating fractal dimensions into our system, incorporating it as an end-to-end task to provide important information on the distributional characteristics of crops and weeds. We evaluated our framework using the BoniRob dataset and the CWFID. When trained with the BoniRob dataset and tested with the CWFID, we obtained a mean intersection of union (mIoU) of 62% and an F1-score of 75.2%. Furthermore, when trained with the CWFID and tested with the BoniRob dataset, we obtained an mIoU of 63.7% and an F1-score of 74.3%. We confirmed that these values are higher than those obtained by state-of-the-art methods.

In the present outbreak of COVID-19, radiographic imaging modalities such as computed tomography (CT) scanners are commonly used for visual assessment of COVID-19 infection. However, personal assessment of CT images is a time-taking process and demands expert radiologists. Recent advancement in artificial intelligence field has achieved remarkable performance of computer-aided diagnosis (CAD) methods. Therefore, various deep learning-driven CAD solutions have been proposed for the automatic diagnosis of COVID-19 infection. However, most of them consider limited number of data samples to develop and validate their methods. In addition, various existing methods employ image-based models considering only spatial information in making a diagnostic decision in case of 3D volumetric data. To address these limitations, we propose a dilated shuffle sequential network (DSS-Net) that considers both spatial and 3D structural features in case of volumetric CT data and makes an effective diagnostic decision. To calculate the performance of the proposed DSS-Net, we combined three publicly accessible datasets that include large number of positive and negative data samples. Finally, our DSS-Net exhibits the average performance of 96.58%, 96.53%, 97.07%, 96.01%, and 98.54% in terms of accuracy, F1-score, average precision, average recall, and area under the curve, respectively, and outperforms various state-of-the-art methods.

Computational cost during model deployment can be reduced through knowledge distillation (KD) which transfers knowledge from a teacher model to a lightweight student model while maintaining performance. However, KD methods based on a single teacher often provide limited knowledge diversity and may transfer less informative features. Furthermore, the application of KD for plant disease classification in low-light noisy environments has received little attention. To address these limitations, we propose a KD framework using heterogeneous teachers with adaptive feature alignment (KDHT-AFA) that integrates an adaptive feature distillation switch (AFDS) to selectively align intermediate representations. For effective knowledge transfer, we propose a lightweight custom student model with 5.16 million parameters. The model incorporates a regional statistical hybrid attention (RSHA) to enhance important features semantically while increasing sensitivity to local structure and contrast. Additionally, we introduce a new self-collected real low-light plant disease dataset (ReLL-PDD v1) acquired under realistic agricultural conditions. For performance evaluation, we considered two open-source datasets, PlantVillage and the Potato Leaf Disease dataset, along with the ReLL-PDD v1 dataset. Our model achieved a mean accuracy of 86.88% with a harmonic mean of precision and recall (F1-score) of 86.25% on ReLL-PDD v1 dataset. On the PlantVillage and Potato Leaf Disease datasets, it obtained mean accuracies of 92.64% and 88.97% with F1-scores of 91.48% and 88.92%, respectively. Furthermore, the smart pesticide prescription module (SPPM) uses a large language model Meta artificial intelligence (LLaMA) to recommend pesticides for diseases detected by KDHT-AFA. Hence, the proposed KDHT-AFA supports real-time deployment and plant disease monitoring in low-light noisy environments.