Figure - available from: Scientific Reports
This content is subject to copyright. Terms and conditions apply.
Study sites used for the sample collection. The map was originally obtained from https://upload.wikimedia.org/wikipedia/commons/a/ab/Thailand_Bangkok_locator_map.svg with license https://creativecommons.org/licenses/by/3.0/deed.en and was modified by using free version of Adobe Illustrator CC 2017 software.
Source publication
Microscopic observation of mosquito species, which is the basis of morphological identification, is a time-consuming and challenging process, particularly owing to the different skills and experience of public health personnel. We present deep learning models based on the well-known you-only-look-once (YOLO) algorithm. This model can be used to sim...
Similar publications
In order to assess risk of mosquito-vector borne disease and to effectively target and monitor vector control efforts, accurate information about mosquito vector population densities is needed. The traditional and still most common approach to this involves the use of traps along with manual counting and classification of mosquito species, but the...
Citations
... Despite their efficiency, the performance of machine learning models can be influenced by dataset quality, feature variety, and environmental unpredictability (Kittichai et al., 2021). Machine learning, especially deep learning models such as CNNs and SVMs, is essential for gender recognition through the analysis of biometric patterns and structural characteristics for automated classification. ...
Gender detection using fingerprint biometrics has emerged as a promising area of research due to its non-intrusive nature and potential applications in biometric identification systems. The procedure can involve multiple steps are the size of finger print and their ridge pattern, minutiae point, machine learning and image processing and accuracy and limitations. This review explores the effectiveness of machine learning techniques for gender classification based on fingerprint patterns, emphasizing the role of advanced classification algorithms and feature extraction methods. Machine learning is crucial for gender detection since it classifies fingerprint patterns and biometric information using models like Convolutional Neural Networks (CNN) and Support Vector Machines (SVM). To identify traits unique to a gender, such as ridge density and minutiae points, these algorithms are trained using labelled datasets. Compared to manual procedures, these models are more effective at handling high-dimensional data and identifying subtle gender-related patterns. Although hybrid models like CNN-DNN and AlexNet further increase classification precision, Convolutional Neural Networks (CNNs) are especially effective due to their automatic feature extraction capabilities. Despite their effectiveness, factors like as picture resolution, demographic balance, and dataset heterogeneity might affect performance, highlighting the need for carefully selected datasets and improved model designs. A structured comparative analysis of multiple studies reveals the impact of various datasets, feature types, and model architectures on classification accuracy and reliability. The findings suggest that deep learning models often outperform traditional classifiers, while dimensionality reduction and hybrid approaches can further enhance performance. However, challenges such as dataset imbalances, limited diversity, and susceptibility to low-quality fingerprint data remain prominent barriers to achieving consistent results. This review also outlines key limitations observed across the studies and provides recommendations for future research, including the need for more diverse datasets and optimized classification frameworks. This study aims to improve fingerprint feature extraction for gender detection, reduce processing costs, fix dataset imbalances, and increase classification accuracy. By stating the objective, the scope and objectives of each investigation are made clear. The generalizability of machine learning models is significantly impacted by the amount, variety, and quality of the dataset. The analysis aims to support the development of more accurate, inclusive, and scalable fingerprint-based gender detection systems.
... This capability has led to impressive performance in various diagnostic tasks, including the analysis of microscopic images [14,17]. A hybrid DL-platform, in particular, has been proposed in several works to help improve the quality and performance of both animal and human malaria detections [17][18][19]. Nonetheless, those techniques are still challenging due to variable of factors, such as limited data of the microscopic image obtained from a cross-sectional time interval of the pathogen life cycle. To improve the training model's performance with greater biological variation of pathological features in a real-world setting, much more open world dataset from other regions is required. ...
... After passing the microscopic images to the welltrained model, the output revealed that animal RBCs were present across every area of the images. Following the detection of each RBC, the rectangular region of each RBC was then cropped utilizing CIRA CORE platform's standard image cropping technique [17,19]. Next, all single-RBC images were then contributed to prepare for the dataset of cropped individual cells as three classes: including (1) the normal RBC as N RBC = 5,300 images, (2) the infected-RBC as N Infected = 5,166 images, and (3) artifact-containing RBC as N Artifact = 1,119 images. ...
... Previous study suggested that the combination of convolutional neural network (CNN) models improved the prediction accuracy [35]. Those research results published learning strategies for identifying genus, species and gender of the mosquito vector [19] and one another for classifying RBC-infected avian malaria [17]. In this study, the YOLOv4 tiny-based object detection is the initial stage of the proposed methodology, with the goal of cropping a single RBC picture from those contained in a microscopic image. ...
Anaplasmosis, which is caused by Anaplasma spp. and transmitted by tick bites, is one of the most serious livestock animal diseases worldwide, causing significant economic losses as well as public health issues. Anaplasma marginale, a gram-negative intracellular obligate bacterium, can cause disease in cattle and other ruminants. Because of the insufficient quality of the slides, a microscopic diagnostic procedure is time-consuming and challenging to diagnose. Intra- and inter-rater variation is frequently imposed on by technicians who are underqualified and unexperienced. Alternatively, algorithms could support local employees in tracking disease transmission and quick action, especially in Thailand where this cattle disease is common. As a result, the study intends to create an automated tool based on a deep neural network linked with an image-retrieval procedure for recognizing infections in microscopic pictures. The Resnext-50 model, which serves as the embedding space’s backbone and is optimized by Triplet-Margin loss, outperforms, with averaged accuracy and specificity ratings of 91.30 percent and 92.83 percent, respectively. The model’s performance was also improved by a fine-tuned procedure between k-nearest neighbor and its normalized distance of each data point, including precision of 0.833 ± 0.134, specificity of 0.930 ± 0.054, recall of 0.838 ± 0.118, and accuracy of 0.915 ± 0.025, respectively. Five-fold cross-validation confirms that the trained model using the optimal k-nearest neighbor (kNN) for the image-based retrieval system, involving 12 images, prevents overfitting via dataset variations indicating areas under the receiver operating curve rankings ranging from 0.917 to 0.922. The image retrieval technique demonstrated in this research is a prototype for a variety of applications. The findings may aid in the early diagnosis of anaplasmosis infections in remote areas without access to veterinary care or costly molecular diagnostic tools.
... In addition, we also compare our proposed model with previous object detection models. Kittichai et al. 24 used two stages of YOLOv2 and v3 to classify 11 species (5 genera) of adult mosquito, achieving a mean average precision of 99% and a sensitivity of 92.4%. Zhao et al. 25 constructed mosquito classification algorithms of 17 species using Swin-transformer. ...
Vector-borne diseases pose a major worldwide health concern, impacting more than 1 billion people globally. Among various blood-feeding arthropods, mosquitoes stand out as the primary carriers of diseases significant in both medical and veterinary fields. Hence, comprehending their distinct role fulfilled by different mosquito types is crucial for efficiently addressing and enhancing control measures against mosquito-transmitted diseases. The conventional method for identifying mosquito species is laborious and requires significant effort to learn. Classification is subsequently carried out by skilled laboratory personnel, rendering the process inherently time-intensive and restricting the task to entomology specialists. Therefore, integrating artificial intelligence with standard taxonomy, such as molecular techniques, is essential for accurate mosquito species identification. Advancement in novel tools with artificial intelligence has challenged the task of developing an automated system for sample collection and identification. This study aims to introduce a self-supervised Vision Transformer supporting an automatic model for classifying mosquitoes found across various regions of Thailand. The objective is to utilize self-distillation with unlabeled data (DINOv2) to develop models on a mobile phone-captured dataset containing 16 species of female mosquitoes, including those known for transmitting malaria and dengue. The DINOv2 model surpassed the ViT baseline model in precision and recall for all mosquito species. When compared on a species-specific level, utilizing the DINOv2 model resulted in reductions in false negatives and false positives, along with enhancements in precision and recall values, in contrast to the baseline model, across all mosquito species. Notably, at least 10 classes exhibited outstanding performance, achieving above precision and recall rates exceeding 90%. Remarkably, when applying cropping techniques to the dataset instead of utilizing the original photographs, there was a significant improvement in performance across all DINOv2 models studied. This is demonstrated by an increase in recall to 87.86%, precision to 91.71%, F1 score to 88.71%, and accuracy to 98.45%, respectively. Malaria mosquito species can be easily distinguished from another genus like Aedes, Mansonia, Armigeres, and Culex, respectively. While classifying malaria vector species presented challenges for the DINOv2 model, utilizing the cropped images enhanced precision by up to 96% for identifying one of the top three malaria vectors in Thailand, Anopheles minimus. A proficiently trained DINOv2 model, coupled with effective data management, can contribute to the development of a mobile phone application. Furthermore, this method shows promise in supporting field professionals who are not entomology experts in effectively addressing pathogens responsible for diseases transmitted by female mosquitoes.
... The Yolov8 model for small pest detection in crops demonstrated the highest mean average precision (mAP) of 84.7%, offering real-time pest detection through an integrated Android application [22]. In mosquito vector surveillance, the YOLOv3-based deep learning models have achieved 99% mean average precision and high sensitivity, aiding in accurate gender and species identification for mosquito populations [23]. The VectorBrain CNN model for malaria vector surveillance achieved 94.4% species classification accuracy and 97.7% sex classification accuracy, providing a promising approach for lightweight, mobile-friendly surveillance in field conditions [24]. ...
... In this case, the extension of image fusion from different body parts needs reevaluation. [26] Gender identification of horsehair crab VGG16 95% F1 score Bjerge et al. [35] Identify nine insects from trap images YOLOv5 92.7% precision Yuan et al. [21] Crop pest detection Improved ResNet34 87.1% Precision Khalid et al. [22] Small pests detection in field crops YOLOv8 84.7% mAP Kittichai et al. [23] Species and gender identification of mosquito YOLOv3 99.0% mAP ...
Sandflies, small insects primarily from the Psychodidae family, are commonly found in sandy, tropical, and subtropical regions. Most active during dawn and dusk, female sandflies feed on blood to facilitate egg production. In doing so, they can transmit infectious diseases that may cause symptoms such as fever, headaches, muscle pain, anemia, skin rashes, and ulcers. Importantly, sandflies are species-specific in their disease transmission. Determining the gender and species of sandflies typically involves examining their morphology and internal anatomy using established identification keys. However, this process requires expert knowledge and is labor-intensive, time-consuming, and prone to misidentification. In this paper, we develop a highly accurate and efficient convolutional network model that utilizes pharyngeal and genital images of sandfly samples to classify the sex and species of three sandfly species (i.e., Phlebotomus sergenti, Ph. alexandri, and Ph. papatasi). A detailed evaluation of the model’s structure and classification performance was conducted using multiple metrics. The results demonstrate an excellent sex-species classification accuracy exceeding 95%. Hence, it is possible to develop automated artificial intelligence-based systems that serve the entomology community at large and specialized professionals.
... This enabled fast scrolling through the video created to quickly detect those frames with mosquitoes and record their landing and departure times. The possibility of analyzing videos using more elaborate image analysis software assisted by machine learning is currently being evaluated [50,51]. ...
Background
Attractive targeted sugar baits (ATSBs) are promising new interventions that can complement existing vector control tools. However, reproducible and quantitative information on the level of attractiveness of ATSBs under field conditions is needed. Therefore, we customized camera traps for close-up imaging. We integrated them into a rugged ATSB monitoring station for day and nighttime recording of mosquitoes landing on the bait.
Methods
The camera traps were evaluated in a semifield system and then in the field in rural Tanzania. In semifield experiments, camera traps were set up in large cages (2 m × 5 m × 2 m) to record mosquitoes landing on an attractive sugar bait (ASB), a blank ASB, or 20% sucrose (w/v). Next, 198 mosquitoes (33 males and 33 females of Anopheles arabiensis, An. funestus and Aedes aegypti) were released into each large cage and allowed to seek a sugar meal for 72 h with a camera recording images of the mosquitoes present on the ASB at 1-min intervals. In the field, 16 camera traps were set in 16 households, 7 with ASB attractant, 7 with ASB blank, and 2 with 20% sucrose (w/v). Human landing catch (HLC) was performed on the same nights as the camera trap recordings.
Results
Under semifield conditions, significantly more mosquitoes visited the ASBs than the blank baits, with An. funestus visiting more frequently than An. arabiensis. There were no significant differences between female and male An. arabiensis visits, but female An. funestus visited more than their conspecific males did. The duration of visits did not vary between the ASB and blank controls or between the mosquito species. Moreover, mosquitoes visited the ASB or sucrose equally, with An. arabiensis visiting the baits more than An. funestus. Compared with male mosquitoes, female mosquitoes visited the baits more often. There was no significant difference in visit duration between the species.
In the field study, a mean of 70 An. arabiensis were caught per person per night on HLC, while 1 individual was caught per night on ASBs. There were significantly more visits by mosquitoes to the ASB than to the ASB blanks or sucrose solution, with more An. arabiensis visiting the baits than An. funestus or Culex quinquefasciatus. Significantly more females than males visited the baits of all the species. Again, the duration of visits was similar among An. arabiensis, An. funestus and C. quinquefasciatus. Aedes aegypti very rarely visited ASBs in the semifield experiments, and none were observed on baits in the field.
Conclusions
Using camera traps to record still images of mosquitoes on ASBs offers reliable, reproducible and quantitative information on their attractiveness in various environmental conditions. Thus, camera traps serve as effective tools for evaluating and improving ATSB technology.
Graphical abstract
... The dry specimens were analyzed microscopically according to the widely accepted taxonomic key. 29 Insectary rearing of mosquitoes. Mosquitoes were reared in an Arthropod Containment Level-2-enabled insectary and maintained at 75 6 5% relative humidity and a set temperature of 27 6 2 C under a strict lightdark cycle (12 hours light and 12 hours dark, including 1-hour dawn/dusk transitions). ...
Mosquitoes are important vectors that transmit viral, protozoan, and helminthic diseases across the world. Climate change and unplanned urbanization are accelerating the spread of these diseases. Controlling vector-borne diseases can be performed most effectively through vector control. Inadequate knowledge of vector bionomics is an impediment and can lead to inappropriate vector control efforts. However, the conventional methods of vector identification are based on morphological differences, demand a significant amount of time and specific skills, and are often misleading. An efficient and affordable solution is needed to quickly and accurately identify pooled samples from vast geographical territories. To ensure the correct identification of distorted or pooled samples in India, a set of definitive steps is required, including the construction of unique primers and the standardization of a one-step assay based on the second internal transcribed spacer gene of the ribosomal DNA. We have successfully developed and confirmed a highly efficient one-step multiplex reverse transcriptase polymerase chain reaction assay for the accurate identification of major mosquito vectors, especially in the cases of both the adult and larval forms of Anopheles sp., Aedes sp., and Culex sp. Hence, the specificity, universality, and uniqueness of these primers could serve as a critical tool for the rapid one-step and one-reaction identification of mosquitoes to control mosquito-borne disease outbreaks and public health emergencies.
... aegypti) and Aedes albopictus (Ae. albopictus) as primary vectors for dengue 61 64 and filariasis 65 . In related work, Eiamsamang, S. conducted experiments comparing mosquito species identification methods between entomologists, public health officers, and deep learning approaches 51 . ...
Traditional mosquito identification methods, relied on microscopic observation and morphological characteristics, often require significant expertise and experience, which can limit their effectiveness. This study introduces a self-supervised learning-based image classification model using the Bootstrap Your Own Latent (BYOL) algorithm, designed to enhance mosquito species identification efficiently. The BYOL algorithm offers a key advantage by eliminating the need for labeled data during pretraining, as it autonomously learns important features. During fine-tuning, the model requires only a small fraction of labeled data to achieve accurate results. Our approach demonstrates impressive performance, achieving over 96.77% accuracy in mosquito image analysis, with minimized both false positives and false negatives. Additionally, the model’s overall accuracy, measured by the area under the ROC curve, surpasses 99.55%, highlighting its robustness and reliability. A notable finding is that fine-tuning with just 10% of labeled data produces results comparable to using the full dataset. This is particularly valuable for resource-limited settings with limited access to advanced equipment and expertise. Our model provides a practical solution for mosquito identification, overcoming the challenges of traditional microscopic methods, such as the time-consuming process and reliance on specialized knowledge in healthcare services. Overall, this model supports personnel in resource-constrained environments by facilitating mosquito vector density analysis and paving the way for future mosquito species identification methodologies.
... From the image detection point of view, most research has focused on adult butterflies but not on more fine-scaled developmental stages, such as instars. The success of CNNs in identifying adult butterflies is partly due to their larger size and often charismatic wings; while many other insects are smaller, or have large variations of size within a species, and in those cases, detection accuracies often fall [27][28][29] . In our case, the body sizes of instars vary significantly, with the first instar (L1), ranging in length from 2 to 6 mm, being much more difficult to localize in photographs compared to the 5th (L5) instar (25-45 mm) (Fig. 1). ...
... One notable challenge of YOLOv5l was ensuring precise instar localization accuracy in the photograph. The model struggles more with images featuring very small instars or those that occupy a minimal area in the photograph, because lower resolution images often present challenges in object (adult mosquito) detection and classification, as highlighted by 28,37 . To explore this, we first calculated the number of pixels within the manually labeled instar bounding box for each image in our hold-out test set, as a way to quantify how "big" or "small" the instar is within an image. ...
Rapid technological advances and growing participation from amateur naturalists have made countless images of insects in their natural habitats available on global web portals. Despite advances in automated species identification, traits like developmental stage or health remain underexplored or manually annotated, with limited focus on automating these features. As a proof-of-concept, we developed a computer vision model utilizing the YOLOv5 algorithm to accurately detect monarch butterfly caterpillars in photographs and classify them into their five developmental stages (instars). The training data were obtained from the iNaturalist portal, and the photographs were first classified and annotated by experts to allow supervised training of models. Our best trained model demonstrates excellent performance on object detection, achieving a mean average precision score of 95% across all five instars. In terms of classification, the YOLOv5l version yielded the best performance, reaching 87% instar classification accuracy for all classes in the test set. Our approach and model show promise in developing detection and classification models for developmental stages for insects, a resource that can be used for large-scale mechanistic studies. These photos hold valuable untapped information, and we’ve released our annotated collection as an open dataset to support replication and expansion of our methods.
... Most models require input images to be of the same size, such as VGG-16, which requires input images to be 224 × 224. 68 Therefore, methods like cropping and resizing are used to adjust images to a uniform size, reducing model training time and computational requirements. 62,69 Normalization can accelerate the learning speed of convolutional neural network (CNN) models and stabilize gradient descent. ...
... The two-stage YOLO v3 achieved the highest average accuracy of 95.6%. 68 For mosquito morphological structure identification, researchers constructed a model based on Mask R-CNN, combined with ResNet101 and Feature Pyramid Network, enhancing the model's ability to learn from images at different scales. This model uses the RPN to automatically map anatomical components of mosquitoes such as thorax, wings, abdomen, and legs from training images. ...
Vector‐borne diseases (VBDs) represent a critical global public health concern, with approximately 80% of the world's population at risk of one or more VBD. Manual disease vector identification is time‐consuming and expert‐dependent, hindering disease control efforts. Deep learning (DL), widely used in image, text, and audio tasks, offers automation potential for disease vector identification. This paper explores the substantial potential of combining DL with disease vector identification. Our aim is to comprehensively summarize the current status of DL in disease vector identification, covering data collection, data preprocessing, model construction, evaluation methods, and applications in identification spanning from species classification to object detection and breeding site identification. We also discuss the challenges and possible prospects for DL in disease vector identification for further research. © 2024 Society of Chemical Industry.
... In 2021, a study used the well-known algorithm "you only_look_once" (YOLO) on 1585 biological samples to recognize different strains of mosquitoes. Thirteen classes were detected with a mean average precision and sensitivity of 99% and 92.4%, respectively (72)(73)(74)(75)(76). By the year 2022, a new algorithm "The Viola-Jones," was designed and used for Leishmania detection. ...
Parasitic infections cause significant morbidity and mortality in tropical and subtropical regions. Accurate and rapid diagnosis is mandatory for effective clinical management. However, the diagnosis of parasitic diseases is defective, especially in developing countries. Researchers have developed new advanced parasitology diagnostics such as smartphones, digital PCR, internet-based bio-surveillance, DNA barcoding, and geometric morphometric analysis. They revealed better sensitivity and specificity, fewer human mistakes, and lower costs. The introduction of artificial intelligence will revolutionize diagnostics when used with these new approaches. Some limitations may be present in developing countries such as internet access and steady Wi-Fi coverage. Hence, combining conventional and advanced methods may decrease limitations and improve diagnosis.
