Identification Task Research Articles

From laboratory to field: cross-domain few-shot learning for crop disease identification in the field

Few-shot learning (FSL) methods have made remarkable progress in the field of plant disease recognition, especially in scenarios with limited available samples. However, current FSL approaches are usually limited to a restrictive setting where base classes and novel classes come from the same domain such as PlantVillage. Consequently, when the model is generalized to new domains (field disease datasets), its performance drops sharply. In this work, we revisit the cross-domain performance of existing FSL methods from both data and model perspectives, aiming to better achieve cross-domain generalization of disease by exploring inter-domain correlations. Specifically, we propose a broader cross-domain few-shot learning(CD-FSL) framework for crop disease identification that allows the classifier to generalize to previously unseen categories and domains. Within this framework, three representative CD-FSL models were developed by integrating the Brownian distance covariance (BCD) module and improving the general feature extractor, namely metric-based CD-FSL(CDFSL-BDC), optimization-based CD-FSL(CDFSL-MAML), and non-meta-learning-based CD-FSL (CDFSL-NML). To capture the impact of domain shift on model performance, six public datasets with inconsistent feature distributions between domains were selected as source domains. We provide a unified testbed to conduct extensive meta-training and meta-testing experiments on the proposed benchmarks to evaluate the generalization performance of CD-FSL in the disease domain. The results showed that the accuracy of the three CD-FSL models improved significantly as the inter-domain similarity increased. Compared with other state-of-the-art CD-FSL models, the CDFSL-BDC models had the best average performance under different domain gaps. Shifting from the pest domain to the crop disease domain, the CDFSL-BDC model achieved an accuracy of 63.95% and 80.13% in the 1-shot/5-shot setting, respectively. Furthermore, extensive evaluation on a multi-domain datasets demonstrated that multi-domain learning exhibits stronger domain transferability compared to single-domain learning when there is a large domain gap between the source and target domain. These comparative results suggest that optimizing the CD-FSL method from a data perspective is highly effective for solving disease identification tasks in field environments. This study holds promise for expanding the application of deep learning techniques in disease detection and provides a technical reference for cross-domain disease detection.

EXPRESS: Exploring the role of visual similarity in parafoveal processing: Insights from the Flanking Letter Lexical Decision Task.

This study explores the impact of visually similar flanking stimuli on central target words using the Flanking Letter Lexical Decision (FLLD) task. Specifically, we investigated whether visual similarity effects can explain orthographic relatedness effects observed in previous FLLD tasks. By employing non-reversal mirror letters as visual flankers, we compared their influence on response times to traditional orthographic-related and orthographic-unrelated conditions. Results confirmed the known facilitative effect of orthographic-related flankers on response times (ROCK ROCK ROCK). However, mirror-related conditions showed no facilitative effect (ROCK ROCK ROCK), as evidenced by a Bayesian analysis indicating no significant differences between mirror-related and mirror-unrelated (STEP ROCK STEP). These findings suggest that low-level visual information in the parafovea does not contribute to the processing of the foveal word in tasks requiring specific word identification. The study concludes that only parafoveal information with relevant linguistic content is spatially pooled across target and flankers during word identification tasks. This research highlights the need to consider task-specific attentional demands and the linguistic relevance of parafoveal information in understanding visual and orthographic processing in reading.

Phoneme identification in bilingual children in Singapore – a preregistered investigation using the CROWN game

ABSTRACT Phoneme identification is critically involved in alphabetic reading, and weak skills in phoneme perception are known to correspond to the development of reading skills in children around the age of reading instruction. However, current models of phoneme perception have been established in monolingual populations, and these relationships may not hold for bilingual children. We developed a phoneme identification task, the CROWN Game, as part of a comprehensive assessment package for identifying potential reading difficulties. In the task, children perform a phoneme identification task, using familiar words with onset consonants spanning a voice-onset time continuum from/b/to/p/(−60 ms to 90 ms). 138 English-Chinese bilinguals in Singapore kindergartens were tested on phoneme perception for English words “beach” and “peach.” We fitted individual responses to a psychometric function and measured individual VOT threshold and the slope of the transition between categories. The task showed a wide spread of scores and good split-half reliability, suggesting that the CROWN game is suitable for identifying stable individual differences between children. In a preregistered subgroup analysis, we examined whether a child’s language background (input ratios and age of acquisition) influenced phoneme perception. GLMMs showed no influence of language exposure on phoneme perception in this task. This suggests that the CROWN Game, as part of a broader assessment package, is valid for use with bilingual children with different rates of exposure to their languages in evaluating phoneme perception skills relevant to reading development.

Modern methods of personal identification

The article presents a review of current scientific research of using digital computer technology in the practical dental identification, analyzes the main problems of their implementation and actual use. Modern aspects of diagnostics and complex planning of identification research are aimed at increasing the efficiency of solving complex medico-legal issues. The application of digital dental technologies has improved the effectiveness of the process of human identification, as well as the reliability of the provided forensic evidence. Significant advantages of the digital photo and X-ray analysis methods in comparison with traditional ones, as well as the advantages of digital facial 3D-reconstruction and dental identification, indicate the potential benefits of using artificial intelligence technology in order to improve the solution of identification tasks. At the same time, data provided by the specialized literature reveal the imperfections of intelligent methods related to discrimination, transparency, accountability, privacy, data security, ethical standards and other important aspects. In this regard, opinions are expressed about the need to partially limit, up to complete rejection of the use of intelligent computer systems to formulate final conclusions and make final decisions on the results of expert studies. In spite of this, leading specialists of the industry are more and more often cautiously declaring the inevitability of virtual evolution of self-developing artificial intelligence systems designed for autonomous existentialism. Referring to the data of modern scientific research, it is worth noting the growing interest of the scientific community in the issues of implementing digital innovative solutions to ensure the effective solution of everyday scientific and practical tasks. Thus, the potential for the successful application of digital technologies to solve the problems of personal identification and improve the quality of forensic evidence provided remains quite promising.

A method of articulating the decision maker's preferences based on examples of decisions made

The paper examines the problem of determining a decision-maker's preference model based on the decisions he makes. This is a fundamental issue in behavioral identification tasks or collaborative information filtering. Inferences about the decision-maker's preferences are made possible by a database containing the values of features accompanying the decisions made. The preference vector serves as the basic preference model, determining the weights of individual coordinates in the decision selection process. The proposed method of preference articulation allows for ordering decisions solely on the basis of the proposed model, without directly referring to the indicated pattern.

Plant Leaf Disease Detection Using Integrated Color and Texture Features

In the realm of precision agriculture, a pivotal challenge lies in the detection, identification, and grading of crop diseases. This multifaceted task necessitates the involvement of expert human resources and time-sensitive actions aimed at mitigating the risks of production losses and the rapid spread of diseases. The effectiveness of the majority of developed systems in this domain hinges on the quality of image features and disease segmentation accuracy. This paper presents a comprehensive research endeavor in the domain of Content-Based Image Retrieval (CBIR), specifically tailored to detect and classify leaf diseases. The proposed system integrates both color and texture features to underpin its functionality, providing a robust framework for accurate disease detection. By leveraging advanced image processing techniques, the system enhances the precision of disease identification, which is crucial for timely and effective intervention in agricultural practices. To evaluate the system’s performance, maize leaves afflicted by rust and blight serve as prime candidates for testing. These diseases were chosen due to their prevalence and significant impact on crop yield. The experimental results demonstrate that the developed system consistently excels in its disease detection and identification tasks, boasting an impressive efficiency rate of 98.33%. This high level of accuracy underscores the potential of the system to be a valuable tool in precision agriculture, aiding farmers and agricultural experts in maintaining healthy crops and optimizing production. The integration of color and texture features not only improves the detection accuracy but also provides a comprehensive understanding of the disease characteristics. This dual-feature approach ensures that the system can distinguish between different types of diseases with high precision, making it a versatile solution for various agricultural applications. The findings of this research highlight the importance of advanced image analysis techniques in enhancing the capabilities of disease detection systems, paving the way for more efficient and effective agricultural practices.

Enhancing image quality using super-resolution residual network for small, blurry images

<span lang="EN-US">In the background, when low-resolution images are utilized, image identification tasks are frequently hampered. By employing the residual network super-resolution framework, super-resolution techniques are used to enhance image quality, specifically in the detection and identification of small and blurry objects. Improving resolution, decreasing blur, and enhancing object detail are the main goals of the suggested approach. The novelty of this research resides in its application of the activation exponential linear unit (ELU) to the super-resolution residual network (SR-ResNet) framework, which has been demonstrated to enhance image sharpness. The experimental findings demonstrate a substantial enhancement in the quality of the images, as evidenced by the training data's structural similarity index (SSIM) of 0.9989 and peak signal-to-noise ratio (PSNR) of 91.8455. Furthermore, the validation data demonstrated SSIM 0.9990 and PSNR 92.5520. The results of this study indicate that the implementation of SR-ResNet significantly enhances the capability of the detection system to detect and classify diminutive and opaque entities precisely. The expected and projected enhancement in image quality significantly influences image processing, especially in situations where accuracy and object differentiation are vital.</span>

Trust with increasing and decreasing reliability.

The primary purpose was to determine how trust changes over time when automation reliability increases or decreases. A secondary purpose was to determine how task-specific self-confidence is associated with trust and reliability level. Both overtrust and undertrust can be detrimental to system performance; therefore, the temporal dynamics of trust with changing reliability level need to be explored. Two experiments used a dominant-color identification task, where automation provided a recommendation to users, with the reliability of the recommendation changing over 300 trials. In Experiment 1, two groups of participants interacted with the system: one group started with a 50% reliable system which increased to 100%, while the other used a system that decreased from 100% to 50%. Experiment 2 included a group where automation reliability increased from 70% to 100%. Trust was initially high in the decreasing group and then declined as reliability level decreased; however, trust also declined in the 50% increasing reliability group. Furthermore, when user self-confidence increased, automation reliability had a greater influence on trust. In Experiment 2, the 70% increasing reliability group showed increased trust in the system. Trust does not always track the reliability of automated systems; in particular, it is difficult for trust to recover once the user has interacted with a low reliability system. This study provides initial evidence into the dynamics of trust for automation that gets better over time suggesting that users should only start interacting with automation when it is sufficiently reliable.

Resilient Semi-Supervised Meta-Learning Network based on wavelet transform and K-means optimization for fluid classification

In the field of geological exploration, accurately distinguishing between different types of fluids is crucial for the development of oil, gas, and mineral resources. Due to the scarcity of labeled samples, traditional supervised learning methods face significant limitations when processing well log data. To address this issue, this paper presents a novel fluid classification method known as the Resilient Semi-Supervised Meta-Learning Network (RSSMLN) based on wavelet transform and K-means optimization, which combines the advantages of few-shot learning and semi-supervised learning, aiming to optimize fluid recognition in well log data. Initially, this study employs a small set of labeled samples to train the initial model and utilizes pseudo-label generation and K-means clustering to optimize prototypes, thereby enhancing the model's accuracy and generalization ability. Subsequently, during the feature extraction process, wavelet transform preprocessing techniques are introduced to enhance the time-frequency feature representation of well log data through multi-scale decomposition. This process effectively captures high-frequency and low-frequency features, providing structured information for subsequent convolution operations. By employing a dual-channel heterogeneous convolutional kernel feature extractor, RSSMLN can effectively capture subtle features of the fluids and significantly improve classification accuracy. Experimental results indicate that compared to various standard deep learning models, RSSMLN achieves superior performance in fluid identification tasks. This research provides a reliable solution for few-shot fluid recognition in oilfield applications and offers scientific support for resource exploration and evaluation.

Joint Classification of Hyperspectral and LiDAR Data via Multiprobability Decision Fusion Method

With the development of sensor technology, the sources of remotely sensed image data for the same region are becoming increasingly diverse. Unlike single-source remote sensing image data, multisource remote sensing image data can provide complementary information for the same feature, promoting its recognition. The effective utilization of remote sensing image data from various sources can enhance the extraction of image features and improve the accuracy of feature recognition. Hyperspectral remote sensing (HSI) data and light detection and ranging (LiDAR) data can provide complementary information from different perspectives and are frequently combined in feature identification tasks. However, the process of joint use suffers from data redundancy, low classification accuracy and high time complexity. To address the aforementioned issues and improve feature recognition in classification tasks, this paper introduces a multiprobability decision fusion (PRDRMF) method for the combined classification of HSI and LiDAR data. First, the original HSI data and LiDAR data are downscaled via the principal component–relative total variation (PRTV) method to remove redundant information. In the multifeature extraction module, the local texture features and spatial features of the image are extracted to consider the local texture and spatial structure of the image data. This is achieved by utilizing the local binary pattern (LBP) and extended multiattribute profile (EMAP) for the two types of data after dimensionality reduction. The four extracted features are subsequently input into the corresponding kernel–extreme learning machine (KELM), which has a simple structure and good classification performance, to obtain four classification probability matrices (CPMs). Finally, the four CPMs are fused via a multiprobability decision fusion method to obtain the optimal classification results. Comparison experiments on four classical HSI and LiDAR datasets demonstrate that the method proposed in this paper achieves high classification performance while reducing the overall time complexity of the method.

Perception of European Portuguese Mid-Vowels by Ukrainian–Russian Bilinguals

Mid-vowel contrasts often present perceptual challenges for speakers of languages that lack these distinctions. However, bilingual speakers, who have access to two phonological systems and exhibit greater metalinguistic awareness, might not necessarily encounter such difficulties. In this study, 27 Ukrainian–Russian bilinguals listened to an unfamiliar language, European Portuguese, and completed two tasks: an identification task where they assimilated the seven stressed oral Portuguese vowels to the closest Ukrainian categories and a discrimination task featuring the Portuguese vowel contrasts /ɛ/–/e/, /e/–/i/, /ɔ/–/o/, and /o/–/u/. No bilingual advantage was observed: the discrimination performance on all contrasts was slightly above or near a chance level (A-prime scores varied between 0.55 and 0.20). These perceptual difficulties may be attributed to the acoustic similarities between the vowels within the contrasts rather than to the differences between the phonological inventories of the languages (the most challenging contrast was not a mid-vowel contrast but acoustically similar /o/–/u/). Although with the back mid-vowel contrast, the difficulty seems to also stem from the possibility that both Ukrainian and Russian have only one back mid-vowel, /o/, and this category occupies a wider area in the vowel space of Ukrainian–Russian bilinguals. The results suggest that bilingual advantage does not always manifest itself in the perception of a new language, especially if two typologically close languages are involved.

Flower Recognition System using CNN

Flower recognition stands as a pivotal task across diverse fields such as botany, agriculture, and environmental studies. The manual process of identifying flower species is not only laborious but also susceptible to inaccuracies. In response to this challenge, this report introduces a Flower Recognition System employing Convolutional Neural Networks (CNNs) to automate and enhance this process. By harnessing the capabilities of deep learning, our system demonstrates remarkable proficiency in accurately classifying various flower species from input images. The CNN architecture, meticulously designed and trained on a comprehensive dataset of labeled flower images, exhibits robustness and efficiency in flower identification tasks. Through rigorous experimentation and evaluation, this report meticulously examines the development and performance of the Flower Recognition System. Furthermore, it explores the potential applications of this system in facilitating automated flower species identification, thereby contributing to advancements in botanical research, agricultural practices, and environmental conservation efforts.

Multimodal feature integration network for lithology identification from point cloud data

Accurate lithology identification from outcrop surfaces is crucial for interpreting geological 3D data. However, challenges arise due to factors such as severe weathering and vegetation coverage, which hinder achieving ideal identification results with both accuracy and efficiency. The integration of 3D point cloud technology and deep learning methodologies presents a promising solution to address these challenges. In this study, we propose a novel multimodal feature integration network designed to distinguish various rock types from point clouds. Our network incorporates a multimodal feature integration block equipped with multiple attention mechanisms to extract representative deep features, along with a hierarchical feature separation block to leverage these features for precise segmentation of points corresponding to different lithologies. Furthermore, we introduce a specialized loss function tailored for rock type identification to enhance network training. Through experiments involving point cloud sampling strategies and loss function evaluation, we identify the optimal network configuration. Comparative analyses against baseline methods demonstrate the superiority of our proposed network across diverse study areas reconstructed from UAV images and laser scanner data, exhibiting improved visual appearance and metric values (Accuracy = 0.978, mean Accuracy = 0.895, mean IoU = 0.857). These findings underscore the efficacy of the multimodal feature integration network as a promising approach for lithology identification tasks in various digital outcrop models derived from heterogeneous data sources.

Antropología forense, conservación y nuevas tecnologías. Aplicación de fotogrametría de rango corto en restos óseos humanos en Córdoba, Argentina

The objective of this work is to contribute to the discussion of the incorporation of new technologies to improve the conservation of human remains, from the implementation of short-range photogrammetry. The study was carried out on human skulls that are under the care of the Institute of Forensic Medicine of the Province of Córdoba, Argentina. Forensic Anthropology and short-range photogrammetry together make up an interdisciplinary field of study where our objective is focused on incorporating new documentation techniques for processing sensible remains and improving the conservation. The incorporation of a digital technique in the practice of this discipline allows us to propose an approach from another perspective to specific problems in this field, allowing the generation of new explanatory possibilities for issues that continually arise during the traditional work of forensic disciplines. In this sense, in the long term, we are interested in generating an information base that in the future will be transformed into a virtual repository with 3D records of human skulls of forensic, bioanthropological, historical and archaeological interest, and in the short term, optimizing the registration and conservation tasks of human remains through their three-dimensional digitization to be able to contribute to the identification and documentation tasks of judicial investigations that enter the institution. On the other hand, we are interested in discussing some ethical concepts focused on the growing digitization of human remains and the care that must be prioritized for their correct diffusion considering that at least in our country there is legislation that protects the exhibition of human remains.

Harnessing slow event-related fMRI to investigate trial-level brain-behavior relationships during object identification.

Understanding brain-behavior relationships is the core goal of cognitive neuroscience. However, these relationships-especially those related to complex cognitive and psychopathological behaviors-have recently been shown to suffer from very small effect sizes (0.1 or less), requiring potentially thousands of participants to yield robust findings. Here, we focus on a much more optimistic case utilizing task-based fMRI and a multi-echo acquisition with trial-level brain-behavior associations measured within participant. In a visual object identification task for which the behavioral measure is response time (RT), we show that while trial-level associations between BOLD and RT can similarly suffer from weak effect sizes, converting these associations to their corresponding group-level effects can yield robust peak effect sizes (Cohen's d = 1.0 or larger). Multi-echo denoising (Multi-Echo ICA or ME-ICA) yields larger effects than optimally combined multi-echo with no denoising, which is in turn an improvement over standard single-echo acquisition. While estimating these brain-behavior relationships benefits from the inclusion of a large number of trials per participant, even a modest number of trials (20-30 or more) yields robust group-level effect sizes, with replicable effects obtainable with relatively standard sample sizes (N = 20-30 participants per sample).

Exploring the relationship among Alzheimer’s disease, aging and cognitive scores through neuroimaging-based approach

Alzheimer’s disease (AD) is a fatal neurodegenerative disorder, with the Mini-Mental State Examination (MMSE) and Clinical Dementia Rating (CDR) serving significant roles in monitoring its progression. We hypothesize that while cognitive assessment scores can detect AD-related brain changes, the targeted brain regions may differ. Additionally, given AD’s strong association with aging, we propose that specific brain regions are influenced by both AD pathology and aging, exhibiting strong correlations with both. To test these hypotheses, we developed a 3D convolutional network with a mixed-attention mechanism to recognize AD subjects from structural magnetic resonance imaging (sMRI) data and utilize 3D convolutional methods to pinpoint brain regions significantly correlated with the AD, MMSE, CDR and age. All models were trained and internally validated on 417 samples from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), and the classification model was externally validated on 382 samples from the Australian Imaging and Lifestyle flagship (AIBL). This approach provided robust support for using MMSE and CDR in assessing AD progression and visually illustrated the relationship between aging and AD. The analysis revealed correlations among the four identification tasks (AD, MMSE, CDR and age) and highlighted asymmetric brain lesions in both AD and aging. Notably, we found that AD can accelerate aging to some extent, and a significant correlation exists between the rate of aging and cognitive assessment scores. This offers new insights into the relationship between AD and aging.

Task-based pronunciation teaching: Lack of auditory precision but not memory hinders learning

This study examined the impact of individual differences in perception and cognition on the effectiveness of task-based pronunciation teaching (TBPT). Seventy young adult Chinese speakers were randomly assigned to either an experimental or control group. The experimental group participated in a 30-min TBPT session, engaging in meaning-oriented tasks designed to help them notice and practice two English vocalic contrasts: [ɛ]-[æ] (e.g., “set” vs. “sat”) and [ɪ]-[i] (e.g., “sit” vs. “seat”). Participants’ learning patterns were then assessed based on their aptitude on perceptual (auditory processing) and cognitive (working memory) levels. Analysis of pre- and post-test results, conducted through forced-choice identification tasks, revealed significant improvements in vowel performance (approximately 10% gain), particularly among participants with normative auditory precision. However, the effectiveness of TBPT was limited among those with lower auditory precision. These findings highlight the negative impact of lower-order aptitude, specifically auditory processing, on the efficacy of TBPT.

Intermediary-guided windowed attention Aggregation network for fine-grained characterization of Major Depressive Disorder fMRI

ObjectivesEstablishing objective and quantitative imaging markers at individual level can assist in accurate diagnosis of Major Depressive Disorder (MDD). However, the clinical heterogeneity of MDD leads to a decrease in recognition accuracy, to address this issue, we propose the Windowed Attention Aggregation Network (WAAN) for a medium-sized functional Magnetic Resonance Imaging (fMRI) dataset comprising 111 MDD and 106 Healthy Controls (HC). MethodsThe proposed WAAN model is a dynamic temporal model that contains two important components, Inner-Window Self-Attention (IWSA) and Cross-Window Self-Attention (CWSA), to characterize the MDD-fMRI data at a fine-grained level and fuse global temporal information. In addition, to optimize WAAN, a new Point to Domain Loss (p2d Loss) function is proposed, which intermediate guides the model to learn class centers with smaller class deviations, thus improving the intra-class feature density. ResultsThe proposed WAAN achieved an accuracy of 83.8 % (±1.4 %) in MDD identification task in medium-sized site. The right superior orbitofrontal gyrus and right superior temporal gyrus (pole) were found to be categorically highly attributable brain regions in MDD patients, and the hippocampus had stable categorical attributions. The effect of temporal parameters on classification was also explored and time window parameters for high categorical attributions were obtained. SignificanceThe proposed WAAN is expected to improve the accuracy of personalized identification of MDD. This study helps to find the target brain regions for treatment or intervention of MDD, and provides better scanning time window parameters for MDD-fMRI analysis.

The role of L1 phonology in the perception of L2 semivowels

Abstract The study compares how L1 Chinese and Japanese speakers perceive L2 English semivowels (/j/ and /w/) preceding high vowels (/i/ and /u/). In Chinese, [j] and [w] serve as phonetic variants of /i/ and /u/, respectively. However, /j/ and /w/ are distinct phonemes in Japanese, although Japanese /w/ lacks the roundness feature found in English /w/. Participants completed experiments with a discrimination task and an identification task. While the discrimination task revealed no differences in /j/ and /w/ perception between the groups, the identification task showed that the Japanese speakers outperformed the Chinese speakers in their perception of /j/, suggesting an L1 “phonemic over phonetic” advantage. However, the Japanese speakers did not outperform the Chinese speakers in their perception of /w/, suggesting that an L2 feature unexploited in the L1 (i.e., roundness) can impede perception. These findings underscore the importance of considering both phonemic status and features in L2 speech perception.

MRNA for Body Fluid and Individual Identification.

Biological stains are one of the most important pieces of evidence, playing a multifaceted role in forensic investigations. An integral facet of forensic practice involves the identification of body fluids, typically achieved through chemical and enzymatic reactions. In recent decades, the introduction of mRNA markers has been posited as a pivotal advancement to augment the capabilities of body fluid identification (BFID). The mRNA coding region single-nucleotide polymorphisms (cSNPs) also present notable advantages, particularly in the task of individual identification. Here, we review the specificity and stability of mRNA markers in the context of BFID and the prowess of mRNA polymorphism in individual identification. Additionally, innovative methods for mRNA detection are discussed.