Figure 1 - available via license: Creative Commons Attribution 4.0 International
Content may be subject to copyright.
Source publication
Unmanned aerial vehicles (UAVs or drones) are a very promising branch of technology, and they have been utilized in agriculture—in cooperation with image processing technologies—for phenotyping and vigor diagnosis. One of the problems in the utilization of UAVs for agricultural purposes is the limitation in flight time. It is necessary to fly at a...
Contexts in source publication
Context 1
... chose tomatoes as our target crop, because tomatoes have the largest number of images (18,149 images) in the dataset. Images of examples of tomato diseases are shown in Figure A1. There are 9 kinds of tomato disease images in the dataset: Xanthomonas campestris pv. ...
Context 2
... neural network design for this research is shown in Figure 1, which was slightly modified from Dong et al. [19] to avoid the image size reduction, and to produce RGB images directly. ...
Context 3
... seems to be caused by the similar appearances of lesion of the diseases: both of Xanthomonas campestris pv. Vesicatoria and Tomato yellow leaf curl virus turn the leaf color into yellow ( Figure A1) On the other hand, in the super-resolution image, the accuracy was improved and almost equal to the accuracy of the high-resolution image. These results indicate that super-resolution method reconstructed the detailed appearance of lesions and enabled the identification of the diseases. ...
Context 4
... this study, we used disease images acquired from ground cameras instead of UAV-based cameras. However, as shown in Figure A1, most of the images in the Plant Village dataset were taken from above a leaf, which are similar to images taken by UAV-based cameras. Although we believe that our approach is effective with UAV-based cameras, it is necessary to evaluate our approach with images taken by UAV-based cameras for practical applications in future studies. ...
Similar publications
Land cover semantic segmentation in high-spatial resolution satellite images plays a vital role in efficient management of land resources, smart agriculture, yield estimation and urban planning. With the recent advancement in remote sensing technologies, such as satellites, drones, UAVs, and airborne vehicles, a large number of high-resolution sate...
RESUMO Com o crescente uso de aeronaves remotamente pilotadas (ARPs) para obtenção de dados de alta resolução espacial, a classificação dessas imagens torna-se uma etapa fundamental para diferentes aplicações de Sensoriamento Remoto. Uma classificação por regiões rigorosa exige uma segmentação adequada que requer a escolha correta de parâmetros, po...
This paper considers the problem of finding a landing spot for a drone in a dense urban environment. The conflicting requirements of fast exploration and high resolution are solved using a multi-resolution approach, by which visual information is collected by the drone at decreasing altitudes so that the spatial resolution of the acquired images in...
Citations
... Yamamoto et al. [51] proposed the super-resolution CNN (SRCNN) with AlexNet architecture on 18,149 PlantVillage images to identify diseases in nine categories, achieving 78.00% accuracy after 50 iterations. In [52], the authors examined the AlexNet and VGG16 pre-trained networks for disease detection across six categories with 13,262 PlantVillage images, where AlexNet achieved 97.49% accuracy after 10 iterations. ...
Recognition of leaf diseases in agriculture is considered a significant aspect of ensuring food quantity, quality, and production. In general, crop leaves are susceptible and fragile to various diseases such as leaf mold, target spot, late blight, bacterial spot or early blight of tomato plants. However, these tomato plant diseases are challenging to recognize, and early diagnosis is vital. At the same time, the continuous growth of convolutional neural network (CNN) approaches has significantly assisted plant disease diagnosis, providing a robust mechanism with highly accurate results. On the other hand, the number of unhealthy leaf images collected is often unbalanced, and diagnosing diseases with such an unbalanced data set is complicated. So, numerous models for tomato disease diagnosis based on CNN models have been proposed. However, none overcomes the class imbalance problem and, as a result, does not generate findings with impartial accuracy. This article presents an efficient and robust solution for the heterogeneous PYNQ-Z1 board. Optimization techniques-including loop unrolling, pipelining, array partitioning, and loop flattening-enhance the computation speed across the network’s convolutional, fully connected, and max-pooling layers. The presented CNN approach comprises an 8-layer network termed MiniTomatoNet. This network is characterized by its streamlined structure, possessing only under 23 K parameters with all weights and biases and occupying a memory of 89.51 KB. In addition, the model trains with a re-weighted focal loss function and achieves 97.63% accuracy and 98.51% AUC score; the inference rate speed is 0.068 s per frame, and the power consumption is 2.35 W. Finally, the model is efficient, low power, robust, high accuracy and fast speed, making it a promising solution for diagnosing tomato diseases.
... Deep learning has become a prominent tool for SR-based plant disease detection. Yamamoto et al. (2017) explored the application of super-resolution techniques to enhance tomato plant disease images, aiming to accelerate image-based phenotyping and vigor diagnosis in agriculture [52]. Cap et al. (2021) proposed LASSR (Locally Adaptive Super-Resolution), specifically designed for plant disease diagnosis, addressing the challenge of artifact suppression in generated images [53]. ...
... Deep learning has become a prominent tool for SR-based plant disease detection. Yamamoto et al. (2017) explored the application of super-resolution techniques to enhance tomato plant disease images, aiming to accelerate image-based phenotyping and vigor diagnosis in agriculture [52]. Cap et al. (2021) proposed LASSR (Locally Adaptive Super-Resolution), specifically designed for plant disease diagnosis, addressing the challenge of artifact suppression in generated images [53]. ...
... These methods only captured channel information from grouped feature maps rather than the original feature maps, resulting these methods incompatible for agricultural images Super-Resolution. Overall, deep learning has fueled rapid progress in image super-resolution [52][53][54][55][56][57][58][59][60][61][62][63][64][65][66][67][68][69][70][71], but key challenges still remain in improving computational efficiency and applying super-resolution to agricultural imaging domain applications like plant leaf images. To overcome these constraints, Adaptive Scale Feature Extraction Super Resolution Network (ASFESRN) is proposed in this paper. ...
Plant diseases pose a significant threat to agricultural productivity, emphasizing the essential need for early detection and diagnosis. While recent deep learning approaches have made strides in plant disease detection, existing methods fall short in accuracy when handling low-resolution (LR) images, often encountered in field conditions due to limited lighting and variable weather. To address this problem, this research introduces a novel Adaptive scale Feature Extraction Super-Resolution Network (ASFESRN) for corn leaf disease detection. The proposed ASFESRN leverages a unique Feature Dismemberment and Integration Block (FDIB) and Adaptive scale Group Convolutional Network (ASGCN) modules, specifically designed to enhance the resolution of LR field images. Unlike previous group convolution methods, the ASGCN module simultaneously extracts spatial and channel information from the original feature map, overcoming limitations and preserving data transmission. The FDIB, incorporating a Channel Attention Module, three ASGCN blocks, and five convolution layers, further enhances the network’s performance in image super resolution. In comparison to existing methods, the ASFESRN demonstrates superior performance in image super resolution, exhibiting improved PSNR, SSIM. The obtained super-resolved (SR) images are then fed into the Disease Detection Block (DDB) for accurate identification of corn leaf diseases. The proposed ASFESRN model underwent extensive training and testing on the PlantVillage Corn Leaf Disease dataset, encompassing various super resolution scaling factors. Results showcase the effectiveness of the proposed model, with accuracies surpassing existing methods for Diverse super resolution scaling factors. The proposed ASFESRN achieves accuracies of 99.7402%, 98.4805%, and 98.9610%, respectively for super resolution scaling factors (𝝬2, 𝝬4, 𝝬6), offering a robust solution for real-time, accurate disease diagnosis in challenging field conditions. This innovative approach equips farmers with early and accurate diagnosis of corn leaf diseases, aiding in effective crop protection.
... Several studies have successfully applied SR techniques to plant images, enhancing the performance of deep learning models in specific tasks. For instance, Yamamoto K et al. [16] deployed the super-resolution of plant disease images for the acceleration of image-based phenotyping and vigor diagnosis in agriculture. Maqsood M H et al. [17] applied superresolution generative adversarial networks [18] for upsampling images before using them to train deep learning models for the detection of wheat yellow rust. ...
Recent advancements in computer vision, especially deep learning models, have shown considerable promise in tasks related to plant image object detection. However, the efficiency of these deep learning models heavily relies on input image quality, with low-resolution images significantly hindering model performance. Therefore, reconstructing high-quality images through specific techniques will help extract features from plant images, thus improving model performance. In this study, we explored the value of super-resolution technology for improving object detection model performance on plant images. Firstly, we built a comprehensive dataset comprising 1030 high-resolution plant images, named the PlantSR dataset. Subsequently, we developed a super-resolution model using the PlantSR dataset and benchmarked it against several state-of-the-art models designed for general image super-resolution tasks. Our proposed model demonstrated superior performance on the PlantSR dataset, indicating its efficacy in enhancing the super-resolution of plant images. Furthermore, we explored the effect of super-resolution on two specific object detection tasks: apple counting and soybean seed counting. By incorporating super-resolution as a pre-processing step, we observed a significant reduction in mean absolute error. Specifically, with the YOLOv7 model employed for apple counting, the mean absolute error decreased from 13.085 to 5.71. Similarly, with the P2PNet-Soy model utilized for soybean seed counting, the mean absolute error decreased from 19.159 to 15.085. These findings underscore the substantial potential of super-resolution technology in improving the performance of object detection models for accurately detecting and counting specific plants from images. The source codes and associated datasets related to this study are available at Github.
... Several studies have successfully applied SR techniques to plant images, enhancing the performance of deep learning models in specific tasks. For instance, Yamamoto K et al. [16] deployed super-resolution of plant disease images for the acceleration of image-based phenotyping and vigor diagnosis in agriculture. Maqsood M H et al. [17] applied super-resolution generative adversarial networks [18] for upsampling the images before using them, to train deep learning models for the detection of wheat yellow rust. ...
Recent advancements in computer vision, especially deep learning models, have shown considerable promise in tasks related to plant image object detection. However, the efficiency of these deep learning models heavily relies on input image quality, with low-resolution images significantly hindering model performance. Therefore, reconstructing high-quality images through specific techniques will help extract features from plant images, thus improve model performance. In this study, we explored the value of super-resolution technology for improving object detection model performance on plant images. Firstly, we built a comprehensive dataset comprising 1030 high-resolution plant images, named the PlantSR dataset. Subsequently, we developed a super-resolution model using the PlantSR dataset and benchmarked it against several state-or-the-art models designed for general image super-resolution tasks. Our proposed model demonstrated superior performance on the PlantSR dataset, indicating its efficacy in enhancing the super-resolution of plant images. Furthermore, we explored the effect of super-resolution on two specific object detection tasks: apple counting and soybean seed counting. By incorporating super-resolution as a pre-processing step, we observed a significant reduction in mean absolute error. Specifically, on the YOLOv7 model employed for apple counting, the mean absolute error decreased from 13.085 to 5.71. Similarly, on the P2PNet-Soy model utilized for soybean seed counting, the mean absolute error decreased from 19.159 to 15.085. These findings underscore the substantial potential of super-resolution technology in improving the performance of object detection models for accurately detecting and counting specific plants from images. The source codes and associated datasets are available at https://github.com/SkyCol/PlantSR.
... Then, the images were color-corrected using Color-Checker Camara Calibration software (X-rite ColorChecker V R Passport; X-rite Inc., Grand Rapids, MI, USA) and Adobe Lightroom (version 6.3; Adobe Systems Inc., San Jose, CA, USA). A method of segmenting the different regions of interest in the image was developed using Image Pro Premier software (version 9.3b; Media Cybernetics, Inc., Rockville, MD, USA) with the smart segmentation tool (Peng et al. 2021;Srinivasagan et al. 2017;Teng et al. 2019;Yamamoto et al. 2017). Images were converted from the red, green, and blue color space to the L* a* b* color space (CIE-International Committee on Illumination, Viena, Austria). ...
During the past 40 years, the US fresh-cut product market has experienced a consistent increase in demand because consumers prioritize health and convenience. Increased interest in fresh-cut products and ready-to-eat vegetables has led to innovations in breeding, product selection, and packaging. However, despite the increased popularity of bell pepper and chile pepper ( Capsicum annuum L.), research of fresh-cut jalapeño pepper is limited. This study was conducted to identify jalapeño cultivars that could be suitable as a raw fresh-cut product and explore measures beyond tissue membrane electrolyte leakage (EL) of processed products that may be useful for the identification of cultivars suitable for fresh-cut applications. A total of 22 fresh-cut parameters were examined across five cultivars of jalapeño peppers and 10 intercrosses of these cultivars, including visual quality based on an image analysis via a computer vision system, package headspace gas composition, tissue membrane EL, and texture. Based on our results, the genotypes were grouped into five clusters using a cluster analysis. Variables including tissue softening (r ² = 0.95), EL (r ² = 0.95), total energy of the mesocarp (r ² = 0.95), and package headspace carbon dioxide (CO 2 ) partial pressure (r ² = 0.94) had strong associations with the cluster. A principal component analysis with biplots further confirmed the results. Cultivars Goliath and Emerald Fire and their hybrids in the first and second clusters showed good quality for fresh-cut applications. The fifth cluster, represented by a single cultivar, Jalapeño M, had the smallest physical size, rapid shelf-life decline, accumulated CO 2 partial pressures, increased EL, and rapid tissue softening in comparison with the other genotypes. All jalapeño cultivars except Jalapeño M maintained good quality until day 14 postprocessing, and some maintained good quality until 21 days postprocessing. Hybrid crosses suggested that two of the cultivars evaluated, Goliath and Emerald Fire, were useful as parents when transferring superior fresh-cut quality traits to progeny. Traditionally, the EL level has been used as an index of freshness (or tissue deterioration). Our results showed that other quality analyses, including measurements of tissue softening via an imaging analysis, and physical analyses of tissue firmness can also be used as indices for the freshness of fresh-cut jalapeños. The results suggest that fruit size, wall thickness, and skin toughness might be useful as predictive measures in the field for the selection of jalapeño genotypes with superior fresh-cut quality.
... It is imperative to acknowledge that the performance of diverse CNN architectures in the context of plant disease identification hinges on a constellation of factors. These include the availability of a limited pool of annotated images, the intricate challenge of accurately representing disease symptoms, nuanced considerations regarding image backgrounds and capturing conditions, and the inherent limitations stemming from the variability in disease symptoms themselves [12]. This multifaceted landscape underscores the complex nature of the task and the dynamic nature of the solutions being developed. ...
Leaf disease detection is a crucial task in modern agriculture, aiding in early diagnosis and prevention of crop infections. In this research paper, authors present a comprehensive study comparing nine widely used pre-trained models, namely DenseNet201, EfficientNetB3, EfficientNetB4, InceptionResNetV2, MobileNetV2, ResNet50, ResNet152, VGG16, and Xception, with our newly developed custom CNN (Convolutional Neural Network) for leaf disease detection. The objective is to determine if our custom CNN can match the performance of these established pre-trained models while maintaining superior efficiency. The authors trained and fine-tuned each pre-trained model and our custom CNN on a large dataset of labeled leaf images, covering various diseases and healthy states. Subsequently, the authors evaluated the models using standard metrics, including accuracy, precision, recall, and F1-score, to gauge their overall performance. Additionally, the authors analyzed computational efficiency regarding training time and memory consumption. Surprisingly, our results indicate that the custom CNN performs comparable to the pre-trained models, despite their sophisticated architectures and extensive pre-training on massive datasets. Moreover, our custom CNN demonstrates superior efficiency, outperforming the pre-trained models regarding training speed and memory requirements. These findings highlight the potential of custom CNN architectures for leaf disease detection tasks, offering a compelling alternative to the commonly used pre-trained models. The efficiency gains achieved by our custom CNN can be beneficial in resource-constrained environments, enabling faster inference and deployment of leaf disease detection systems. Overall, our research contributes to the advancement of agricultural technology by presenting a robust and efficient solution for the early detection of leaf diseases, thereby aiding in crop protection and yield enhancement.
... For instance, ML approaches such as Support Vector Machines, Random Forest, Decision Trees, Naïve Bayes, and K-Nearest Neighbours have been deployed for the identification of diseased samples using images of bell-pepper (Anjna et al., 2020), maize (Panigrahi et al., 2020), tomato (Agarwal et al., 2020;Harakannanavar et al., 2022), and rice (Shrivastava & Pradhan, 2021;Zamani et al., 2022). Further, application of more advanced computational tools such as deep-learning has greatly improved information processing for plant health assessment following data acquisition via machine vision (Ghosal et al., 2018;Nagasubramanian et al., 2019;Yamamoto et al., 2017). A wide variety of deep-learning algorithms implementing unique iterations of convolutional neural networks have also been successfully tested for plant disease detection in various crops, such as cassava (Sambasivam & Opiyo, 2021), tomato (Abbas et al., 2021;Agarwal et al., 2020;Chowdhury et al., 2021;Harakannanavar et al., 2022), maize (Li et al., 2020), peach (Bedi & Gole, 2021), and strawberry (Shin et al., 2021). ...
Use of vertical farms is increasing rapidly as it enables year-round crop production, made possible by fully controlled growing environments situated within supply chains. However, intensive planting and high relative humidity make such systems ideal for the proliferation of fungal pathogens. Thus, despite the use of bio-fungicides and enhanced biosecurity measures, contamination of crops does happen, leading to extensive crop loss, necessitating the use of high-throughput monitoring for early detection of infected plants. In the present study, progression of foliar symptoms caused by Pythium irregulare -induced root rot was monitored for flat-leaf parsley grown in an experimental hydroponic vertical farming setup. Structural and spectral changes in plant canopy were recorded non-invasively at regular intervals using a 3D multispectral scanner. Five morphometric and nine spectral features were selected, and different combinations of these features were subjected to multivariate data analysis via principal component analysis to identify temporal trends for early segregation of healthy and infected samples. Combining morphometric and spectral features enabled a clear distinction between healthy and diseased plants at 4–7 days post inoculation (DPI), whereas use of only morphometric or spectral features allowed this at 7–9 DPI. Minimal datasets combining the six most effective features also resulted in effective grouping of healthy and diseased plants at 4–7 DPI. This suggests that selectively combining morphometric and spectral features can enable accurate early identification of infected plants, thus creating the scope for improving high-throughput crop monitoring in vertical farms.
... Moreover, high-resolution sensors and high-end computers can lead to expensive hardware expenses. However, levering software capabilities such as optimized deep learning algorithms trained with high-resolution data can significantly reduce hardware costs while preserving high-quality data acquisition in new precision agriculture solutions [19][20][21]. ...
In the context of recent technological advancements driven by distributed work and open-source resources, computer vision stands out as an innovative force, transforming how machines interact with and comprehend the visual world around us. This work conceives, designs, implements, and operates a computer vision and artificial intelligence method for object detection with integrated depth estimation. With applications ranging from autonomous fruit-harvesting systems to phenotyping tasks, the proposed Depth Object Detector (DOD) is trained and evaluated using the Microsoft Common Objects in Context dataset and the MinneApple dataset for object and fruit detection, respectively. The DOD is benchmarked against current state-of-the-art models. The results demonstrate the proposed method’s efficiency for operation on embedded systems, with a favorable balance between accuracy and speed, making it well suited for real-time applications on edge devices in the context of the Internet of things.
... This section primarily focuses on the study conducted by multiple scholars in the field of leaf disease detection via deep learning methodologies. Drone-captured images of tomato disease were analysed using a super resolution-CNN technique by Kyosuke Yamamota et al. [20] to recover finer details such as lesion on plant organ. Researchers from all over the world presented their findings in 2018, including Barbedo et al. ...
In recent years, numerous deep learning architectures have used publicly available/author-generated datasets to classify plant diseases. This study suggested a four-stage process for classifying plant diseases using a comparative evaluation method based on deep learning. First, the ideal CNN was found by comparing basic CNN designs with adapted and hybrid versions of contemporary DL models. Second, we tried to train the best acquired model through Adam deep learning optimizers to increase its performance. Xception and VGG-16 architecture were compared using a number of different performance criteria. These metrics included validation accuracy/loss, F1-score, precision, recall, AuC and RoC. All of the chosen DL architectures were trained on three leaf species such as tomato, potato and bell pepper of the PlantVillage dataset, which includes information about 26 unique diseases that have been seen in 14 distinct plant species. To train deep learning architectures, we used Keras with TensorFlow as the backend. Overall, the Adam optimizer-trained VGG-16 architecture performed best in terms of training and validation accuracy of (98.42%, 100%, 99.75%) and (91.85%, 97.21% and 98.18%) for dataset-1, dataset-2 and dataset-3 respectively, demonstrating its superiority over prior methods and establishing its unique contribution to the field. This study's proposed method can, therefore, be used in other agricultural applications for detection and categorization in a more open and accurate manner.
... Tarek et al., [63] ...
All people have access to food thanks to agriculture, even in areas with rapid population expansion. In order to ensure that the entire population has access to food, early diagnosis of plant diseases is recommended. However, when crops are still maturing, it might be challenging to predict diseases. The goal of this paper is to inform farmers about the use of machine learning to recognize various pest species in various crop plants. Deep learning establishes a continuing, cutting-edge method for analysing images with significant potential and promising outcomes. DL has expanded into the field of agriculture after demonstrating its effectiveness in a number of applications. Here, we reviewed various research articles that used deep learning methods to address diverse research issues in tomato (Solanum lycopersicum) plants. We look at the study areas in tomato plants where deep learning, data preprocessing, transfer learning, and augmentation approaches are performed. studied dataset details, including the number of photos, classes, and train test to validation ratios that were employed. We also look at comparisons made across different deep learning architectures and analyse the results. The results demonstrated that deep learning techniques beat all other image processing methods, however how well DL works depended on the dataset employed.
