No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Transparent Intelligent Vision for Black Sigatoka Detection
ResearchGate has not been able to resolve any citations for this publication.
Stem rust is a highly prevalent and damaging fungal disease that affects wheat crops worldwide. It is caused by the Puccinia graminis f. sp. tritici fungus, and can significantly reduce wheat production and quality, by 10% to 50%. In severe cases, it causes up to a 90% decrease in grain yield if left uncontrolled. Manual diagnosis by pathologists and experts which use visual inspection is usually costly, time-consuming, and prone to errors. Moreover, the scarcity of pathologists in rural areas poses a challenge in detecting and grading wheat stem rust. To address these issues, this study presents a system for automated detection and classification of wheat stem rust based on the modified Cobbs scale (CIMMYT) guidelines. The system employs image pre-processing techniques such as adaptive thresholding for segmentation and the Gabor filter for feature extraction which enhances textural features to identify key disease characteristics. For disease classification, a deep learning approach was utilized which employs a 12-way Softmax to assign specific classes: Resistant (tR), Moderately resistant (MR), and Susceptible (S). The proposed model is trained and tested using an image dataset collected in collaboration with pathologists from Haramaya University. The proposed model achieves a training accuracy of 92.02% and a testing accuracy of 92.01% in grading wheat stem rust, outperforming state-of-the-art models such as CNN. The study's main achievements are achieved in the areas of real-time monitoring, and grading precision. Real-time monitoring provides timely updates on disease presence and severity, enabling proactive management and minimizing crop losses. Grading precision offers a quantitative assessment of disease damage, facilitating targeted treatment strategies and resource allocation.
The rapid development of artificial intelligence (AI) is leading to major philosophical and ethical questions about the nature of intelligence, the relationship between humans and machines, and the future of humanity. The philosophy of AI is particularly critical in ensuring the responsible and ethical development and use of AI in Society 5.0, a future society where technology and AI play a central role in improving the quality of life and solving complex problems. More precisely, the philosophies of AI today should be concerned with transparency, accountability, privacy, and preventing discrimination and bias in AI algorithms. This chapter provides a foundation for understanding the impact of AI on humanity and guides the development of AI technologies to align with society’s values and aspirations. It examines the philosophies of AI with their specific interplays in the role of shaping the future of Society 5.0, ensuring that AI systems contribute to a better future for all. By understanding these philosophical underpinnings, researchers and AI practitioners can create responsible AI methods and evaluation approaches that conform to reponsible data standards and AI practices globally.
Visualisation techniques are powerful tools to understand the behaviour of Artificial Intelligence (AI) systems. They can be used to identify important features contributing to the network decisions, investigate biases in datasets, and find weaknesses in the system's structure (e.g., network architectures). Lawmakers and regulators may not allow the use of smart systems if these systems cannot explain the logic underlying a decision or action taken. These systems are required to offer a high level of 'transparency' to be approved for deployment. Model transparency is vital for safety-critical applications such as autonomous navigation and operation systems (e.g., autonomous trains or cars), where prediction errors may have serious implications. Thus, being highly accurate without explaining the basis of their performance is not enough to satisfy regulatory requirements. The lack of system interpretability is a major obstacle to the wider adoption of AI in safety-critical applications. Explainable Artificial Intelligence (XAI) techniques applied to intelligent systems to justify their decisions offers a possible solution. In this review, we present state-of-the-art explanation techniques in detail. We focus our presentation and critical discussion on visualisation methods for the most adopted architecture in use, the Convolutional Neural Networks (CNNs), applied to the domain of image classification. Further, we discuss the evaluation techniques for different explanation methods, which shows that some of the most visually appealing methods are unreliable and can be considered a simple feature or edge detector. In contrast, robust methods can give insights into the model behaviour, which helps to enhance the model performance and boost the confidence in the model's predictions. Besides, the applications of XAI techniques show their importance in many fields such as medicine and industry. We hope that this review proves a valuable contribution for researchers in the field of XAI.
Maintenance decisions in domains such as aeronautics are becoming increasingly dependent on being able to predict the failure of components and systems. When data-driven techniques are used for this prognostic task, they often face headwinds due to their perceived lack of interpretability. To address this issue, this paper examines how features used in a data-driven prognostic approach correlate with established metrics of monotonicity, trendability, and prognosability. In particular, we use the SHAP model (SHapley Additive exPlanations) from the field of eXplainable Artificial Intelligence (XAI) to analyze the outcome of three increasingly complex algorithms: Linear Regression, Multi-Layer Perceptron, and Echo State Network. Our goal is to test the hypothesis that the prognostics metrics correlate with the SHAP model's explanations, i.e., the SHAP values. We use baseline data from a standard data set that contains several hundred run-to-failure trajectories for jet engines. The results indicate that SHAP values track very closely with these metrics with differences observed between the models that support the assertion that model complexity is a significant factor to consider when explainability is a consideration in prognostics.
This paper investigates a deep learning method in image classification for the detection of colorectal cancer with ResNet architecture. The exceptional performance of a deep learning classification incites scholars to implement them in medical images. In this study, we trained ResNet-18 and ResNet-50 on colon glands images. The models trained to distinguish colorectal cancer into benign and malignant. We assessed our prototypes on three varieties of testing data (20%, 25%, and 40% of whole datasets). The empirical outcomes confirm that the application of ResNet-50 provides the most reliable performance for accuracy, sensitivity, and specificity value than ResNet-18 in three kinds of testing data. Upon three test assortments, we perceive the best performance value on 20% and 25% test sets with a classification accuracy of above 80%, the sensitivity of above 87%, and the specificity of above 83%. In this research, a deep learning method demonstrates the profoundly reliable and reproducible outcomes for biomedical image analysis.
Deep learning (DL) represents the golden era in the machine learning (ML) domain, and it has gradually become the leading approach in many fields. It is currently playing a vital role in the early detection and classification of plant diseases. The use of ML techniques in this field is viewed as having brought considerable improvement in cultivation productivity sectors, particularly with the recent emergence of DL, which seems to have increased accuracy levels. Recently, many DL architectures have been implemented accompanying visualisation techniques that are essential for determining symptoms and classifying plant diseases. This review investigates and analyses the most recent methods, developed over three years leading up to 2020, for training, augmentation, feature fusion and extraction, recognising and counting crops, and detecting plant diseases, including how these methods can be harnessed to feed deep classifiers and their effects on classifier accuracy.
While Black Sigatoka Leaf Disease (Mycosphaerella fijiensis) has arguably been the most important pathogen affecting the banana industry over the past 50 years, there are no quantitative estimates of what risk factors determine its spread across the globe, nor how its spread has affected banana producing countries. This study empirically models the disease spread across and its impact within countries using historical spread timelines, biophysical models, local climate data, and country level agricultural data. To model the global spread a empirical hazard model is employed. The results show that the most important factor affecting first time infection of a country is the extent of their agricultural imports, having increased first time disease incidence by 69% points. In contrast, long distance dispersal due to climatic factors only raised this probability by 0.8% points. The impact of disease diffusion within countries once they are infected is modelled using a panel regression estimator. Findings indicate that under the right climate conditions the impact of Black Sigatoka Leaf Disease can be substantial, currently resulting in an average 3% reduction in global annual production, i.e., a loss of yearly revenue of about USD 1.6 billion.
Food security for the 7 billion people on earth requires minimizing crop damage by timely detection of diseases. Most deep learning models for automated detection of diseases in plants suffer from the fatal flaw that once tested on independent data, their performance drops significantly. This work investigates a potential solution to this problem by using segmented image data to train the convolutional neural network (CNN) models. As compared to the F-CNN model trained using full images, S-CNN model trained using segmented images more than doubles in performance to 98.6% accuracy when tested on independent data previously unseen by the models even with 10 disease classes. Not only this, by using tomato plant and target spot disease type as an example, we show that the confidence of self-classification for S-CNN model improves significantly over F-CNN model. This research work brings applicability of automated methods closer to non-experts for timely detection of diseases.
Sigatoka leaf diseases are a major constraint to banana production. A survey was conducted in Tanzania and Uganda to assess the distribution of Pseudocercospora species and severity of Sigatoka leaf diseases. Pseudocercospora species were identified using species‐specific primers. Sigatoka‐like leaf diseases were observed in all farms and on all cultivars, but disease severity varied significantly (p<0.001) between countries, districts/regions within countries, altitudinal ranges and banana cultivars. In all regions except Kilimanjaro, P. fijiensis, the causal agent of black Sigatoka, was the only pathogen associated with Sigatoka disease. Mycosphaerella musae was associated with Sigatoka‐like symptoms in Kilimanjaro region. Black Sigatoka disease was more severe in Uganda with a mean disease severity index (DSI) of 37.5% than Tanzania (DSI = 19.9%). In Uganda, black Sigatoka disease was equally severe in Luwero district (mean DSI = 40.4%) and Mbarara district (mean DSI = 37.9%). In Tanzania, black Sigatoka was most severe in Kagera region (mean DSI = 29.2%) and least in Mbeya region (mean DSI = 11.5%). P. fijiensis was detected at altitudes of up to 1877 m. The range expansion of the most devastating sigatoka pathogen, P. fijiensis, previously confined to altitudes less than 1350 m in East Africa is worrisome, especially for smallholder banana farmers growing the susceptible East African Highland Bananas (EAHB). Among the banana varieties sampled, the EAHB, FHIA hybrids and Mchare were the most susceptible. We report, for the first time the loss of resistance in Yangambi KM5; a banana variety previously resistant to P. fijiensis.
Deep learning techniques, and in particular Convolutional Neural Networks (CNNs), have led to significant progress in image processing. Since 2016, many applications for the automatic identification of crop diseases have been developed. These applications could serve as a basis for the development of expertise assistance or automatic screening tools. Such tools could contribute to more sustainable agricultural practices and greater food production security. To assess the potential of these networks for such applications, we survey 19 studies that relied on CNNs to automatically identify crop diseases. We describe their profiles, their main implementation aspects and their performance. Our survey allows us to identify the major issues and shortcomings of works in this research area. We also provide guidelines to improve the use of CNNs in operational contexts as well as some directions for future research.
Chest X-Rays (CXRs) are widely used for diagnosing abnormalities in the heart and lung area. Automatically detecting these abnormalities with high accuracy could greatly enhance real world diagnosis processes. Lack of standard publicly available dataset and benchmark studies, however, makes it difficult to compare various detection methods. In order to overcome these difficulties, we have used a publicly available Indiana CXR, JSRT and Shenzhen dataset and studied the performance of known deep convolutional network (DCN) architectures on different abnormalities. We find that the same DCN architecture doesn't perform well across all abnormalities. Shallow features or earlier layers consistently provide higher detection accuracy compared to deep features. We have also found ensemble models to improve classification significantly compared to single model. Combining these insight, we report the highest accuracy on chest X-Ray abnormality detection on these datasets. We find that for cardiomegaly detection, the deep learning method improves the accuracy by a staggering 17 percentage point compared to rule based methods. We applied the techniques to the problem of tuberculosis detection on a different dataset and achieved the highest accuracy. Our localization experiments using these trained classifiers show that for spatially spread out abnormalities like cardiomegaly and pulmonary edema, the network can localize the abnormalities successfully most of the time. One remarkable result of the cardiomegaly localization is that the heart and its surrounding region is most responsible for cardiomegaly detection, in contrast to the rule based models where the ratio of heart and lung area is used as the measure. We believe that through deep learning based classification and localization, we will discover many more interesting features in medical image diagnosis that are not considered traditionally.
Black Sigatoka or black leaf streak disease, caused by the Dothideomycete fungus Pseudocercospora fijiensis (previously: Mycosphaerella fijiensis), is the most significant foliar disease of banana worldwide. Due to the lack of effective host resistance, management of this disease requires frequent fungicide applications, which greatly increase the economic and environmental costs to produce banana. Weekly applications in most banana plantations lead to rapid evolution of fungicide-resistant strains within populations causing disease-control failures throughout the world. Given its extremely high economic importance, two strains of P. fijiensis were sequenced and assembled with the aid of a new genetic linkage map. The 74-Mb genome of P. fijiensis is massively expanded by LTR retrotransposons, making it the largest genome within the Dothideomycetes. Melting-curve assays suggest that the genomes of two closely related members of the Sigatoka disease complex, P. eumusae and P. musae, also are expanded. Electrophoretic karyotyping and analyses of molecular markers in P. fijiensis field populations showed chromosome-length polymorphisms and high genetic diversity. Genetic differentiation was also detected using neutral markers, suggesting strong selection with limited gene flow at the studied geographic scale. Frequencies of fungicide resistance in fungicide-treated plantations were much higher than those in untreated wild-type P. fijiensis populations. A homologue of the Cladosporium fulvum Avr4 effector, PfAvr4, was identified in the P. fijiensis genome. Infiltration of the purified PfAVR4 protein into leaves of the resistant banana variety Calcutta 4 resulted in a hypersensitive-like response. This result suggests that Calcutta 4 could carry an unknown resistance gene recognizing PfAVR4. Besides adding to our understanding of the overall Dothideomycete genome structures, the P. fijiensis genome will aid in developing fungicide treatment schedules to combat this pathogen and in improving the efficiency of banana breeding programs.
Most existing bottom-up methods measure the foreground saliency of a pixel or region based on its contrast within a local context or the entire image, whereas a few methods focus on segmenting out background regions and thereby salient objects. Instead of considering the contrast between the salient objects and their surrounding regions, we consider both foreground and background cues in a different way. We rank the similarity of the image elements (pixels or regions) with foreground cues or background cues via graph-based manifold ranking. The saliency of the image elements is defined based on their relevances to the given seeds or queries. We represent the image as a close-loop graph with super pixels as nodes. These nodes are ranked based on the similarity to background and foreground queries, based on affinity matrices. Saliency detection is carried out in a two-stage scheme to extract background regions and foreground salient objects efficiently. Experimental results on two large benchmark databases demonstrate the proposed method performs well when against the state-of-the-art methods in terms of accuracy and speed. We also create a more difficult benchmark database containing 5,172 images to test the proposed saliency model and make this database publicly available with this paper for further studies in the saliency field.
Plantains and bananas are staple food for about 70 million people throughout the humid and sub-humid tropic of Africa. They provide an important source of revenue for small holders who cultivate them. Production, however, has always been threatened by a variety of constraints; the most overriding constraint being black Sigatoka disease caused by a wind-borne fungus, Mycosphaerella fijiensis Morelet. As much as 27% of the total cost of production is apportioned to curb the menace the disease. Hence in this write up, the various methods applied so far to control black sigatoka disease in plantains and bananas are reviewed with emphasis on their apparent challenges and prospects. Findings showed that the consumption of plantains and bananas has risen tremendously in recent years, and black Sigatoka disease can be controlled in various ways – culturally, chemically, quarantine, and breeding for disease resistance. A proper management of organic matter using different crop residues as mulch builds up the soil fertility level, and substantially reduced the effect of the disease. The use of forecasting methods could be part of an integrated disease management strategy, as this would reduce the number of fungicide treatment, disease production cost, and partially eliminate pollution challenges. The production and cultivation of disease resistant cultivars in combination with good cultural practices is generally considered to be the most appropriate intervention strategies that would control black sigatoka disease of plantains and bananas.
Explainability has become one of the most discussed topics in machine learning research in recent years, and although a lot of methodologies that try to provide explanations to black-box models have been proposed to address such an issue, little discussion has been made on the pre-processing steps involving the pipeline of development of machine learning solutions, such as feature selection. In this work, we evaluate a game-theoretic approach used to explain the output of any machine learning model, SHAP, as a feature selection mechanism. In the experiments, we show that besides being able to explain the decisions of a model, it achieves better results than three commonly used feature selection algorithms.
Composed in honour of the sixty-fifth birthday of Lloyd Shapley, this volume makes accessible the large body of work that has grown out of Shapley's seminal 1953 paper. Each of the twenty essays concerns some aspect of the Shapley value. Three of the chapters are reprints of the 'ancestral' papers: Chapter 2 is Shapley's original 1953 paper defining the value; Chapter 3 is the 1954 paper by Shapley and Shubik applying the value to voting models; and chapter 19 is Shapley's 1969 paper defining a value for games without transferable utility. The other seventeen chapters were contributed especially for this volume. The first chapter introduces the subject and the other essays in the volume, and contains a brief account of a few of Shapley's other major contributions to game theory. The other chapters cover the reformulations, interpretations and generalizations that have been inspired by the Shapley value, and its applications to the study of coalition formulation, to the organization of large markets, to problems of cost allocation, and to the study of games in which utility is not transferable.
The Nelson Mandela African Institution of Science and Technology Bananas dataset
- N Mduma
Diseases of banana, abaca and enset
- D R Jones
A holistic integrated management approach to control black sigatoka disease of banana caused by mycosphaerella fijiensis
- L Pérez-Vicente