Automatic greenhouse insect pest detection and recognition based on a cascaded deep learning classification method

Inspection of insect sticky paper traps is an essential task for an effective integrated pest management (IPM) programme. However, identification and counting of the insect pests stuck on the traps is a very cumbersome task. Therefore, an efficient approach is needed to alleviate the problem and to provide timely information on insect pests. In this research, an automatic method for the multi-class recognition of small-size greenhouse insect pests on sticky paper trap images acquired by wireless imaging devices is proposed. The developed algorithm features a cascaded approach that uses a convolutional neural network (CNN) object detector and CNN image classifiers, separately. The object detector was trained for detecting objects in an image, and a CNN classifier was applied to further filter out non-insect objects from the detected objects in the first stage. The obtained insect objects were then further classified into flies (Diptera: Drosophilidae), gnats (Diptera: Sciaridae), thrips (Thysanoptera: Thripidae) and whiteflies (Hemiptera: Aleyrodidae), using a multi-class CNN classifier in the second stage. Advantages of this approach include flexibility in adding more classes to the multi-class insect classifier and sample control strategies to improve classification performance. The algorithm was developed and tested for images taken by multiple wireless imaging devices installed in several greenhouses under natural and variable lighting environments. Based on the testing results from long-term experiments in greenhouses, it was found that the algorithm could achieve average F₁-scores of 0.92 and 0.90 and mean counting accuracies of 0.91 and 0.90, as tested on a separate 6-month image data set and on an image data set from a different greenhouse, respectively. The proposed method in this research resolves important problems for the automated recognition of insect pests and provides instantaneous information of insect pest occurrences in greenhouses, which offers vast potential for developing more efficient IPM strategies in agriculture.     

An artificial neural network approach to identify fungal diseases of cucumber (Cucumis sativus L.) plants using digital image processing

Nowadays, artificial intelligence solutions such as digital image processing and artificial neural networks (ANN) have become important applicable techniques in phytomonitoring and plant health detection systems. In this research, an autonomous device was designed and developed for detecting two types of fungi (Pseudoperonospora cubensis, Sphaerotheca fuliginea) that infect the cucumber (Cucumis sativus L.) plant leaves. This device was able to recognise the fungal diseases of plants by detecting their symptoms on plant leaves (downy mildew and powdery mildew). For leaves of cucumber inoculated with different spores of the fungi, it was possible to estimate the amount of hour post inoculation (HPI) by extracting leaves’ image parameters. Device included a dark chamber, a CCD digital camera, a thermal camera, a light dependent resistor lightening module and a personal computer. The proposed programme for precise disease detection was based on an image processing algorithm and ANN. Three textural features and two thermal parameters from the obtained images were measured and normalised. Performance of ANN model was tested successfully for disease recognition and detecting HPI in images using back-propagation supervised learning method and inspection data. Such this machine vision system can be used in robotic intelligent systems to achieve a modern farmer’s assistant in agricultural crop fields.     

Application of an image and environmental sensor network for automated greenhouse insect pest monitoring

This work presents an automated insect pest counting and environmental condition monitoring system using integrated camera modules and an embedded system as the sensor node in a wireless sensor network. The sensor node can be used to simultaneously acquire images of sticky paper traps and measure temperature, humidity, and light intensity levels in a greenhouse. An image processing algorithm was applied to automatically detect and count insect pests on an insect sticky trap with 93% average temporal detection accuracy compared with manual counting. The integrated monitoring system was implemented with multiple sensor nodes in a greenhouse and experiments were performed to test the system’s performance. Experimental results show that the automatic counting of the monitoring system is comparable with manual counting, and the insect pest count information can be continuously and effectively recorded. Information on insect pest concentrations were further analyzed temporally and spatially with environmental factors. Analyses of experimental data reveal that the normalized hourly increase in the insect pest count appears to be associated with the change in light intensity, temperature, and relative humidity. With the proposed system, laborious manual counting can be circumvented and timely assessment of insect pest and environmental information can be achieved. The system also offers an efficient tool for long-term insect pest behavior observations, as well as for practical applications in integrated pest management (IPM).     

New image dataset and new negative sample judgment method for crop pest recognition based on deep learning models

Crop pests are responsible for serious economic loss around the worldwide. Accurate recognition of pests is the key to pest control and is a considerable challenge in farming. Deep learning models have shown great promise in image recognition, drawing the attention of many agricultural experts. However, the lack of pest image datasets and the inexplicability of deep learning models have hindered the development of deep learning models in the field of pest recognition. Our work provides the following four contributions: (1) We constructed a new and more effective dataset, for crop pest recognition, named IP41 comprising 46,567 original images of crop pests in 41 classes. (2) We trained three different deep learning models based on IP41, using transfer learning combined with fine-tuning. The results of the three deep learning models exceeded 80.00% recognition. (3) A negative sample judgment method was proposed to exclude the uploaded pest-free images of the user. (4) We provided reasonable visual explanations for the most critical areas of the recognition layers by using the gradient-weighted class activation mapping method. This research suggests that the recognition process focuses more on image details than the image as a whole, and that overall difference is ignored to a certain extent. These results will be helpful to future research in the field of agricultural pest recognition     

Chromosome-level genome assembly of an agricultural pest,the rice leaffolder Cnaphalocrocis exigua (Crambidae,Lepidoptera)

The rice leaffolder Cnaphalocrocis exigua (Crambidae, Lepidoptera) is an important agricultural pest that damages rice crops and other members of related grass families. C. exigua exhibits a very similar morphological phenotype and feeding behaviour to C. medinalis, another species of rice leaffolder whose genome was recently reported. However, genomic information for C. exigua remains extremely limited. Here, we used a hybrid strategy combining different sequencing technologies, including Illumina, PacBio, 10× Genomics, and Hi – C scaffolding, to generate a high-quality chromosome-level genome assembly of C. exigua. We initially obtained a 798.8 Mb assembly with a contig N50 size of 2.9 Mb, and the N50 size was subsequently increased to 25.7 Mb using Hi – C technology to anchor 1413 scaffolds to 32 chromosomes. We detected a total of 97.7% Benchmarking Universal Single-Copy Orthologues (BUSCO) in the genome assembly, which was comprised of ~52% repetitive sequence and annotated 14,922 protein-coding genes. Of note, the Z and W sex chromosomes were assembled and identified. A comparative genomic analysis demonstrated that despite the high synteny observed between the two rice leaffolders, the species have distinct genomic features associated with expansion and contraction of gene families and selection pressure. In summary, our chromosome-level genome assembly and comparative genomic analysis of C. exigua provide novel insights into the evolution and ecology of this rice insect pests and offer useful information for pest control.     

A single semi‐synthetic diet with improved antimicrobial activity for mass rearing of lepidopteran insect pests of cotton and maize

Helicoverpa armigera (Hübner), Earias vittella (Fabricius), Spodoptera litura (Fabricius), Spodoptera exigua (Hübner) (all Lepidoptera: Noctuidae), Pectinophora gossypiella (Saunders) (Lepidoptera: Gelechiidae), and Chilo partellus (Swinhoe) (Lepidoptera: Crambidae) are the major pests of cotton and maize. Mass‐rearing of these insects under controlled conditions is necessary to obtain the numbers needed to conduct bioassays to screen insecticides, proteins, and other compounds, as tools for insect pest management. We present a diet suitable for rearing the six lepidopteran pests (five cotton and one maize pest). We further show that this diet is on par with or superior to the published diet recipes for each of the insect species, which were studied for three generations. We also discuss the advantages of antimicrobials other than formalin for keeping microbial growth under check. A combination of antimicrobial solution and benomyl provided effective control and suppressed the growth of microbes for a longer period than a formalin‐containing diet. A common diet for six pests provide opportunities for automation of diet preparation in addition to improved throughput and consistency in the process, while eliminating diet‐batch related errors.     

Non-specific resistance in brown planthopper resistant indica rice varieties against Plodia interpunctella (Lepidoptera: Phycitidae)

Mechanisms of host plant resistance against insect pests can be manifold. Resistance screenings generally use single target insect pests, but the resistance thus screened may not always be specific to the target insect species. We conducted a test for non‐specific resistance in indica rice varieties with resistance genes against brown planthopper (BPH), by using the Indian meal moth, Plodia interpunctella. The test system was very simple, and only required the non‐pest moth to be reared on rice flour. We compared the survival rate, developmental period and adult weight of the moth on three rice varieties: ‘Nipponbare’, a BPH‐susceptible japonica variety, and ‘Thai Collection 11’ and ‘Pokkali’, two resistant indica varieties. Our results were straightforward and demonstrate that resistance in the two resistant rice varieties is not BPH specific, because development of the moth was retarded and adult body weight was reduced.     

Aspects of insect chemical ecology: exploitation of reception and detection as tools for deception of pests and beneficial insects

Empirical exploitation of insect reception and detection at the peripheral neurosensory level has been extremely valuable for identifying pheromones and other semiochemicals, mainly by electroantennogram or single cell preparations coupled with capillary gas chromatography. Differential sensitivity to semiochemicals at the single‐cell level has allowed the identification of some of the most active semiochemicals relating to host location and, more importantly, to the avoidance of nonhosts. However, in terms of molecular recognition, there is still a considerable gap in understanding the detection of particular molecules and their discrimination from closely‐related chemical structures. New approaches will be needed to understand the processes of molecular recognition more precisely. Nevertheless, from electrophysiological studies to the most advanced molecular techniques, it has been possible to identify semiochemicals for the deception of pests in their quest to find plant and animal hosts, as well as mates. Even the deception of insects antagonistic to pests, particularly parasitoids, can now be exploited for managing pests in more sustainable systems. Successes in exploiting insect semiochemicals in the interests of better agriculture and animal husbandry are exemplified, and potential new ways of learning more about reception and detection for deception are discussed. This takes the subject beyond the management of pest and beneficial insects to wider commercial and social opportunities.     

基于性诱和深度学习的草地贪夜蛾成虫自动识别计数方法

【目的】探究深度学习在草地贪夜蛾Spodoptera frugiperda成虫自动识别计数上的可行性,并评估模型的识别计数准确率,为害虫机器智能监测提供图像识别与计数方法。【方法】设计一种基于性诱的害虫图像监测装置,定时自动采集诱捕到的草地贪夜蛾成虫图像,结合采集船形诱捕器粘虫板上草地贪夜蛾成虫图像,构建数据集;应用YOLOv5深度学习目标检测模型进行特征学习,通过草地贪夜蛾原始图像、清除边缘残缺目标、增加相似检测目标(斜纹夜蛾成虫)、无检测目标负样本等不同处理的数据集进行模型训练,得到Yolov5s-A1, Yolov5s-A2, Yolov5s-AB, Yolov5s-ABC 4个模型,对比在不同遮挡程度梯度下的测试样本不同模型检测结果,用准确率(P)、召回率(R)、F1值、平均准确率(average precision, AP)和计数准确率(counting accuracy, CA)评估各模型的差异。【结果】通过原始图像集训练的模型Yolov5s-A1的识别准确率为87.37%,召回率为90.24%,F1值为88.78;清除边缘残缺目标图像集训练得到的模型Yolov5s-A2的识别准确率为93.15%,召回率为84.77%,F1值为88.76;增加斜纹夜蛾成虫样本图像训练的模型Yolov5s-AB的识别准确率为96.23%,召回率为91.85%,F1值为93.99;增加斜纹夜蛾成虫和无检测对象负样本训练的模型Yolov5s-ABC的识别准确率为94.76%,召回率为88.23%,F1值为91.38。4个模型的AP值从高到低排列如下：Yolov5s-AB>Yolov5s-ABC> Yolov5s-A2>Yolov5s-A1,其中Yolov5s-AB与Yolov5s-ABC结果相近;CA值从高到低排列如下：Yolov5s-AB>Yolov5s-ABC>Yolov5s-A2>Yolov5s-A1。【结论】结果表明本文提出的方法应用于控制条件下害虫图像监测设备及诱捕器粘虫板上草地贪夜蛾成虫的识别计数是可行的,深度学习技术对于草地贪夜蛾成虫的识别和计数是有效的。基于深度学习的草地贪夜蛾成虫自动识别与计数方法对虫体姿态变化、杂物干扰等有较好的鲁棒性,可从各种虫体姿态及破损虫体中自动统计出草地贪夜蛾成虫的数量,在害虫种群监测中具有广阔的应用前景。     

Current status and future perspectives on natural enemies for pest control in Taiwan

ABSTRACT

In Taiwan, the agricultural policy, ‘Reduce the consumption of pesticide to half in the next 10 years’, was launched in 2017. Pesticide application, which results in contamination of food by chemical residues, pest resistance, and other adverse ecological effects, is a growing public and environmental concern. Pest control by natural predators is, thus, the best alternative. Biological control methods implemented based on insights obtained from studies on pest behaviour, rearing, and various crop management modes, increase the possibility of controlling pests in modern organic agricultural systems. More than a decade has passed since the first introduction of a predatory insect in Taiwan for pest control (in the 1990s). Predatory and parasitic natural enemies, including lacewing, predatory stink bugs, Orius, and parasitic wasps, were initially used for controlling thrips, aphids, spider mites, whiteflies, and lepidopteran pests. At present, there exists a wide range of integrated pest management (IPM) methods incorporating other non-chemical, biological, and agricultural methods. However, recently, there has been an increase in research and development on the utilisation of natural enemies of insects and the associated food safety issues. Mass production and release, storage, and handling techniques of insect predators and parasitoids have been successful in recent years. The final goal of present day research is to develop natural enemy products and provide an IPM-based model to farmers for using natural enemies in agricultural production systems, thereby reducing pesticide application and ensuring food security.     

Switching among natal and auxiliary hosts increases vulnerability of Spodoptera exigua (Hübner) (Lepidoptera: Noctuidae) to insecticides

The role of insecticidal application and host plant resistance in managing Spodoptera exigua has been well documented, but the effect of different host plants, on which the pest cycles its population in the field, has seldom been investigated. Therefore, we have studied the vulnerability of S. exigua against commonly used insecticides (cypermethrin, chlorpyrifos, lufenuron, and emamectin benzoate) with different mode of actions when it switches its generations from natal to auxiliary hosts and vice versa. Different field populations being established on different host plants including castor, cauliflower, cotton, okra, and spinach were collected and reared in the laboratory before insecticidal bioassays. The role of larval diet and host plant switching on their response to tolerate applied insecticides was studied using leaf‐dip bioassay methods. Host switching demonstrated a significant role in altering the vulnerability of S. exigua populations to tested insecticides. Spodoptera exigua sourced from castor, when switched host to okra and spinach, exhibited 50% higher mortality when treated with emamectin benzoate. This trend in mortality was consistent upon complete host switch cycle (natal—auxiliary—natal host). However, the highest increase (92%) in vulnerability was recorded when the larvae were shifted to spinach from cotton. In general, chlorpyrifos and lufenuron had highest efficacies in terms of larval mortality. The findings of present studies provide insights to a better understanding the behavior of polyphagous pests and the role of different host plants in altering the susceptibility of these pests against applied insecticides. Ultimately the results warrant that due consideration should be given to cropping patterns and time of host switching by pest population during planning and executing chemical control.     

昆虫数学形态学研究及其应用展望

数学形态学是用数学方法描述或分析一个物体图象的形状的理论和方法,是图象处理和图象识别技术的发展,但在生物学当中的应用还很有限。本文介绍了一个新的分支学科——昆虫数学形态学,包括三方面的内容：①昆虫数学形态学技术研究,涉及昆虫图象数字化技术和昆虫图象处理与识别技术;②昆虫数学形态学理论研究,主要以昆虫图象的解释和理解研究及昆虫数学形态学与分类学等学科的关系研究为主;③昆虫和昆虫数学形态学应用基础研究,涉及昆虫数学形态学数据库及其分析软件开发,昆虫图象的机器学习和计算机视觉等内容。昆虫数学形态学理论和方法与计算机视觉技术相结合,在害虫虫情监测、昆虫多媒体专家系统的构建等方面具有广阔的应用前景。     

转Bt基因抗虫棉的生态风险及治理对策

评述了转Bt基因抗虫棉的生态风险及治理对策。其生态风险主要表现在目标害虫的抗性和对非目标生物群落的变化。目标害虫与转基因抗虫棉的互相作用和抗虫棉杀虫毒素的时空表达方式是目标害虫抗性发展的主要途径。在转基因抗虫棉田中,虽然对目标害虫的防治次数大为减少,但害虫和天敌群落的稳定性仍不如常规棉田,某种次要害虫大发生的可能性较大。认为将转基因抗虫棉纳入综合防治体系并培育更加高效的抗虫棉是治理目标害虫抗性和防止次要害虫上升的重要措施。     

Imaging techniques for the detection of stored product pests

Stored grains are subject to deterioration and losses through various factors, but mainly insects and fungi. Various techniques are employed to detect stored product pests; however, there is an urgent need for an industrial-scale on-line detection technique. Near-infrared hyperspectroscopic imaging and soft X-rays have shown the potential for real-time application. These techniques are particularly effective for detecting internal infestations of stored grains. The digital images of the scanned objects are analyzed for various spectral and image features using statistical techniques such as complex multivariate tools. Classification accuracies as high as 80–100 % have been achieved for various pest and grain combinations. Dual-energy X-rays have been shown to detect the concealed eggs of stored product insect pests. The main threats to stored cereals come from Aspergillus spp., Penicillium spp., and Fusarium spp., which may produce mycotoxins. These imaging techniques have shown good results in the detection of fungal infections of stored grain.     

CNN and transformer framework for insect pest classification

Insect pests pose a significant and increasing threat to agricultural production worldwide. However, most existing recognition methods are built upon well-known convolutional neural networks, which limits the possibility of improving pest recognition accuracies. This research attempts to overcome this challenge from a novel perspective, constructing a simplified but very useful network for effective insect pest recognition by combining transformer architecture and convolution blocks. First, the representative features are extracted from the input image using a backbone convolutional neural network. Second, a new transformer attention-based classification head is proposed to sufficiently utilize spatial data from the features. With that, we explore different combinations for each module in our model and abstract our model into a simple and scalable architecture; we introduce more effective training strategies, pretrained models and data augmentation methods. Our models performance was evaluated on the IP102 benchmark dataset and achieved classification accuracies of 74.897% and 75.583% with minimal implementation costs at image resolutions of 224 × 224 pixels and 480 × 480 pixels, respectively. Our model also attains accuracies of 99.472% and 97.935% on the D0 dataset and Li's dataset, respectively, with an image resolution of 224 × 224 pixels. The experimental results demonstrate that our method is superior to the state-of-the-art methods on these datasets. Accordingly, the proposed model can be deployed in practice and provides additional insights into the related research.     

Bioecology of some key cashew insect pests and diseases in diverse habitats and landscapes in Tanzania

Cashew (Anacardium occidentale) is an economically important cash crop for many rural households in Tanzania. However, its production is constrained by some insect pests and diseases. As a prerequisite for the development of a more sustainable integrated insect pest and disease management strategy for cashew, information on the biology and ecology of the key insect pests and diseases in a changing environment, and on influencing biotic and abiotic factors, is needed. Surveys were conducted in the major cashew nut‐producing areas of Tanzania for two seasons: August to December, 2009, and August to December, 2010. Data on number of infested and infected shoots by key insect pests and diseases, natural enemies and associated farmer practices, namely synthetic pesticide use and intercropping systems, were collected from different subzones within agroecological zones. Our data showed that abundance and diversity of key cashew insect pests and diseases were influenced by agroecological zones and subzones. Intercropping was more commonly practised in the northern than in the southern zone. Agrochemicals were most frequently used in the southern agroecological zone and affected the occurrence of natural enemies, notably the weaver ant that was more abundant in the northern zone. Furthermore, our findings revealed that Helopeltis sp. and the powdery mildew remained the major constraints to cashew nut production in Tanzania.     

Genetic control of Plutella xylostella in omics era

Diamondback moth, Plutella xylostella (L.), is a specialist pest on cruciferous crops of economic importance. The large‐scale use of chemical insecticides for the control of this insect pest has caused a number of challenges to agro‐ecosystems. With the advent of the omics era, genetic pest management strategies are becoming increasingly feasible and show a powerful potential for pest control. Here, we review strategies for using transgenic plants and sterile insect techniques for genetic pest management and introduce the major advances in the control of P. xylostella using a female‐specific RIDL (release of insects carrying a dominant lethal gene) strategy. Further, the advantages of gene drive developed in combination with sex determination and CRISPR/Cas9 systems are addressed, and the corresponding prospects and implementation issues are discussed. It is predictable that under the policy and regulation of professional committees, the genetic pest control strategy, especially for gene drive, will open a new avenue to sustainable pest management not only for P. xylostella but also for other insect pests.     

Predicting insect pest status under climate change scenarios: combining experimental data and population dynamics modelling

Climate change could profoundly affect the status of agricultural insect pests. Several approaches have been used to predict how the temperature and precipitation changes could modify the abundances, distributions or status of insect pests. In this article it is demonstrated how the use of simple models, such as Ricker’s classic equation, including a mechanistic representation of the influence of exogenous forces may improve our predictive capacity of the dynamic behaviour of insect populations. Using data from classical experiments in population ecology, we evaluate how temperature and humidity influence the density of two stored grain insect pest, Tribolium confusum and Callosobruchus chinensis, and then, using the A2 and B2 scenarios proposed by the Intergovernmental Panel on Climate Change and the previous modelling, we develop predictions over the future pest status of T. confusum along South America austral region, and specifically for eight cities in the continental Chilean territory. Tribolium confusum and C. chinensis show qualitatively different responses to the exogenous forcing of temperature and humidity, respectively. Our simulations predict a change in the equilibrium density of T. confusum from 10 to 14% under the moderate B2 scenario and 12 to 22% under the extreme A2 scenario to the period, 2071–2100. Both results imply a severe change in the pest status of this species in the southern region. This study illustrates how the use of theoretically based models may improve our predictive capacity. This approach provides an opportunity to examine the link between invasive species and climate change and how new suitable habitat may become available for species whose niche space is limited in some degree by climatic conditions. The use of different scenarios allows us to examine the sensitivity of the predictions, and to improve the communication with the general public and decision‐makers; a key aspect in integrated pest management.     

害虫灾害研究的复杂性理论框架

害虫灾害是高度复杂的大系统 ,表现出不均匀性、差异性、多样性、突发性、随机性、可预测性和周期性等复杂性特征 ,使得经典的理论和方法已不适用于害虫灾害的研究。依据复杂性科学和分形、神经网络、混沌及小波等非线性科学的发展及其近期在害虫灾害中的部分研究成果 ,该文从复杂大系统出发 ,构建了害虫灾害研究的复杂性理论框架 ,为深入研究害虫灾害的成因、机制与预测提供理论依据。