Recognition Challenge Research Articles

A Vehicle Re-Identification Algorithm for Long-Distance Small Targets Combining YOLOv8 Object Detection Algorithm

With the rapid development of intelligent technologies, deep learning-based object detection has been widely applied in dynamic fields, especially in vehicle management. However, challenges in accurate recognition and tracking remain, particularly due to issues like intra-class variations and inter-class similarities in vehicle re-identification across camera viewpoints. This paper proposes a method for long-range small vehicle target re-identification based on the YOLOv8 object detection algorithm, aiming to improve the detection accuracy and re-identification performance of small targets in complex environments. First, a detection and re-identification dataset containing long-range vehicle images is constructed based on a complex background and small target dataset. Secondly, the YOLOv8-EMA-RFB optimization algorithm is introduced, which combines the EMA (Exponential Moving Average) attention mechanism and RFB (Receptive Field Block) structure. The EMA module enhances feature extraction for small targets and reduces noise interference, while the RFB module increases the receptive field, improving the detection capability for small targets. Through these optimizations, the proposed method effectively improves the detection accuracy of long-range small targets in the YOLOv8 model. Additionally, by incorporating the AlignedReID method, feature matching in object re-identification is improved, further enhancing the accuracy and robustness of multi-object tracking. Experimental results demonstrate that the proposed method achieves high precision and real-time performance in multi-scene vehicle detection and re-identification, offering a novel solution for intelligent transportation and urban security surveillance.

Researching on insulator defect recognition based on context cluster CenterNet++

The existing UAV inspection images are faced with many challenges for insulator defect recognition. A new multi-resolution Context Cluster CenterNet++ model is proposed. First, this paper proposes the Context Cluster method to solve the problem of low recognition accuracy caused by non-uniform distribution of targets. The cluster region is used to identify and predict the location of the target, and the improved loss function is used to modify the cluster center. Secondly, this paper uses deformable convolution operator (DCNv2) combined with path aggregation network (PAN) to carry out deformable convolution operation on the image, and accurately predicts the regression box and key point triplet (KP), so as to improve the accurate positioning of the target position of any shape and any scale. The sensitivity of the model to target scale change and deformation is reduced, and the recognition accuracy of the model is improved. Then, Bhattacharyya distance is used to calculate the triplet prediction loss of key points and the target center point offset loss, so as to significantly improve the positioning accuracy of the same target in different frames. Finally, experiments are carried out on the MS-COCO dataset and the National Grid standardized UAV inspection insulator image dataset. Our code is at https://github.com/mengbonannan88/CC-CenterNet.

The challenges of recognition among the Tjwa people in Zimbabwe

For more than a century, the primordial inhabitants of the Zimbabwe, the Tjwa people, have experienced numerous injustices. The most significant of these has been the conspicuous absence of recognition from both the colonial and post-colonial administrations. It was not until the 2013 Constitution, in Chapter 1, Section 6 (1), that the Tjwao language – erroneously categorised as Khoisan – was finally acknowledged by the current government. Tragically, this indigenous language is now on the brink of extinction, with fewer than 3000 members, and less than 20 fluent speakers in the Tsholotsho region. This paper explores the challenges faced by these people using qualitative research methods, which reinforced their marginalisation in the context of Zimbabwe, in terms of culture and language.

Advancing Emotion Recognition in Wearable Devices: Core Technologies, Challenges and Practical Solutions

This paper reviews the advancements and challenges in emotion recognition systems, focusing on their applications in wearable devices. It examines core technologiesmultimodal data fusion, deep learning models, and personalized frameworksin addressing critical issues such as data quality, computational efficiency, and privacy protection. By highlighting innovative solutions, including edge computing, lightweight architectures, and advanced privacy-preserving techniques, the paper emphasizes their role in enhancing the robustness, scalability, and security. By integrating real-world applications with state-of-the-art methodologies, this review underscores the transformative potential of wearable emotion recognition technologies for enhancing well-being and daily life. This review also examines current challenges in emotion recognition and provides potential solutions to address these issues

YOLOv8n-CA: Improved YOLOv8n Model for Tomato Fruit Recognition at Different Stages of Ripeness

This study addresses the challenges of tomato maturity recognition in natural environments, such as occlusion caused by branches and leaves, and the difficulty in detecting stacked fruits. To overcome these issues, we propose a novel YOLOv8n-CA method for tomato maturity recognition, which defines four maturity stages: unripe, turning color, turning ripe, and fully ripe. The model is based on the YOLOv8n architecture, incorporating the coordinate attention (CA) mechanism into the backbone network to enhance the model’s ability to capture and express features of the tomato fruits. Additionally, the C2f-FN structure was utilized in both the backbone and neck networks to strengthen the model’s capacity to extract maturity-related features. The CARAFE up-sampling operator was integrated to expand the receptive field for improved feature fusion. Finally, the SIoU loss function was used to solve the problem of insufficient CIoU of the original loss function. Experimental results showed that the YOLOv8n-CA model had a parameter count of only 2.45 × 106, computational complexity of 6.9 GFLOPs, and a weight file size of just 4.90 MB. The model achieved a mean average precision (mAP) of 97.3%. Compared to the YOLOv8n model, it reduced the model size slightly while improving accuracy by 1.3 percentage points. When compared to seven other models—Faster R-CNN, YOLOv3s, YOLOv5s, YOLOv5m, YOLOv7, YOLOv8n, YOLOv10s, and YOLOv11n—the YOLOv8n-CA model was the smallest in size and demonstrated superior detection performance.

Seatizen Atlas: a collaborative dataset of underwater and aerial marine imagery

Citizen Science initiatives have a worldwide impact on environmental research by providing data at a global scale and high resolution. Mapping marine biodiversity remains a key challenge to which citizen initiatives can contribute. Here we describe a dataset made of both underwater and aerial imagery collected in shallow tropical coastal areas by using various low cost platforms operated either by citizens or researchers. This dataset is regularly updated and contains >1.6 M images from the Southwest Indian Ocean. Most of images are geolocated, and some are annotated with 51 distinct classes (e.g. fauna, and habitats) to train AI models. The quality of these photos taken by action cameras along the trajectories of different platforms, is highly heterogeneous (due to varying speed, depth, turbidity, and perspectives) and well reflects the challenges of underwater image recognition. Data discovery and access rely on DOI assignment while data interoperability and reuse is ensured by complying with widely used community standards. The open-source data workflow is provided to ease contributions from anyone collecting pictures.

Human Identification System using Images

The present study focuses on the challenges of human long-range recognition with deep learning, and AlexNet performance is considered, being one of the most recognized convolutional neural networks. A specially curated dataset was used to mimic the real world in a more practical and realistic scenario, which involves subjects varying in distances and angles while creating low-resolution issues with variability in lighting and distractions from the environment. The system starts by passing the input dataset through a local Laplacian filter that is utilized to enhance the quality of images before feeding it into a pre-trained AlexNet model. The results reveal that AlexNet reaches a recognition accuracy of 65% and shows that image properties such as sharpness and contrast significantly affect performance, as measured by the mean and standard deviation. The study underlines requirements of high-quality images for applications in long-range human recognition and sheds insight into the ways enhancements in images could improve performance in deep learning models. Results support further evolution of more potent human recognition systems, particularly in application domains such as surveillance and security in which the accurate identification of a target from a long distance is highly critical. Key Words: Long-range recognition, Human recognition, Deep learning, Image processing

The Effect of Contralateral Routing of Signal Devices on the Quality of Life of Unilateral Cochlear Implant Recipients and Their Frequent Communication Partners.

Unilateral cochlear implant (CI) recipients with limited hearing in the contralateral ear are deprived of the advantages of binaural hearing. To address speech recognition challenges arising from the head shadow effect, a contralateral routing of signal (CROS) device can be used; however, less is known of the broader impact of a CROS device on an individual's quality of life (QoL) or that of their frequent communication partners (FCPs). This preliminary study aimed to evaluate the impact of CROS on speech recognition in noise and its influence on the QoL of unilateral CI recipients and their FCPs. This preliminary study enrolled seven adult unilateral CI recipients and their FCPs. All CI recipients were fitted with CROS devices during their initial appointments. Speech recognition testing was conducted in noise with and without the CROS device in a sound booth before a take-home trial. Participants used the CROS devices for approximately 1 year, with device fitting occurring before and continuing during the COVID-19 pandemic. Participants completed two QoL questionnaires, the Auditory Performance and Satisfaction Scale for Single-Sided Deafness (APS-SSD) and the Nijmegen Cochlear Implant Questionnaire (NCIQ), twice: once prior to CROS device use and once after the take-home trial. Additionally, the FCPs of each CI recipient completed the Significant Other Scale of Hearing Disability (SOS-HEAR) Questionnaire twice, once before and once after extended CROS device use. When noise was directed toward the CI ear and speech toward the non-CI ear, speech recognition improved by 32% with the CROS device (p = .001). CI recipients reported significant median improvement in the "general" domain of the APS-SSD after the take-home trial (Wilcoxon Z = 12.0, p < .05). FCPs reported a significant median reduction in concerns related to their partner's hearing when the CI recipient used the CROS device (Wilcoxon Z = 2.0, p < .05). This preliminary study demonstrates the benefit of CROS devices for unilateral CI recipients in noisy environments. Additionally, it highlights the positive impact of CROS devices on the QoL of both CI recipients and their FCPs. These findings emphasize the importance of considering CROS devices as a valuable solution for unilateral CI recipients to enhance their hearing experience, overall well-being, and that of their FCPs.

Neural correlates of facial recognition deficits in autism spectrum disorder: a comprehensive review.

Autism spectrum disorder (ASD) is a neurodevelopmental condition characterized by significant impairments in social interaction, often manifested in facial recognition deficits. These deficits hinder individuals with ASD from recognizing facial identities and interpreting emotions, further complicating social communication. This review explores the neural mechanisms underlying these deficits, focusing on both functional anomalies and anatomical differences in key brain regions such as the fusiform gyrus (FG), amygdala, superior temporal sulcus (STS), and prefrontal cortex (PFC). It has been found that the reduced activation in the FG and atypical activation of the amygdala and STS contribute to difficulties in processing facial cues, while increased reliance on the PFC for facial recognition tasks imposes a cognitive load. Additionally, disrupted functional and structural connectivity between these regions further exacerbates facial recognition challenges. Future research should emphasize longitudinal, multimodal neuroimaging approaches to better understand developmental trajectories and design personalized interventions, leveraging AI and machine learning to optimize therapeutic outcomes for individuals with ASD.

Advancements and Challenges in Character Recognition: A Comparative Analysis of CNN and Deep Learning Approaches

Advancements and Challenges in Character Recognition: A Comparative Analysis of CNN and Deep Learning Approaches

Listening to The Reluctant Fundamentalist:

This article centers on Mohsin Hamid's The Reluctant Fundamentalist (2007) and explores the challenges of reciprocal recognition and the impact of power dynamics on trust between the West and Muslim communities in post 9/11 America. Bart Moore-Gilbert's critique of existing models of recognition politics; specifically, models developed by Fukuyama and Taylor, reveals the limitations in addressing the Muslim experience and Islamic fundamentalism within western liberal-democratic societies. The failure of ocular-centric approaches – prevalent in the western tradition – to adequately represent the Muslim experience, guides me towards sound theories which provides the theoretical underpinning to this analysis. Thus, I denote this shift as a practice in decolonial listening. Drawing on Nicole Furlonge’s analytical framework of “listen in print” and the concept of “aural recognition”, I argue for a more comprehensive recognition through an aurally-engaged practice of reading and political engagement. Ultimately, I advocate for listening in order to develop trust and for a compassionate stance towards lives and encounters of Muslim minorities in post 9/11 America.

Reading Patterns in L2 Dyslexic Italian Learners of Chinese: An Eye Tracking Perspective

The study investigates the reading patterns of dyslexic learners of Chinese through the lens of eye-tracking technology, focusing on the differences in cognitive processing when engaging with two types of Chinese characters: ideo-phonetic compounds and pictographs.Dyslexia, a specific learning disorder, significantly impacts reading abilities, leading to challenges in word recognition, spelling, and comprehension. The research aims to enhance understanding of how visual stimuli can aid dyslexic students in learning Chinese, a language characterized by its unique writing system.The study involved ten Italian university students, including six diagnosed with dyslexia, who participated in a 100-hour online Chinese language course. Eye movements were recorded using a Tobii Pro Spectrum eye tracker, allowing for an analysis of fixation patterns and durations as participants engaged with the characters. The research questions focused on whether the type of character influenced the average number and duration of fixations, and whether there were significant differences in fixation patterns between dyslexic and non-dyslexic readers.Results indicated that both dyslexic and non-dyslexic participants exhibited longer fixation durations on pictographs compared to ideo-phonetic compounds, suggesting a more complex cognitive processing for pictographic characters. However, contrary to initial hypotheses, dyslexic participants did not demonstrate a greater number of fixations on ideo-phonetic compounds. The findings highlight the importance of understanding character structures and suggest that knowledge of basic character components may be more beneficial for dyslexic learners than visual aids alone.This research represents a significant step in exploring the intersection of dyslexia and Chinese language acquisition in Italy, providing empirical data that can inform educational strategies tailored for dyslexic students. The study concludes that while visual stimuli can support learning, a deeper understanding of character structures is crucial for improving reading comprehension among dyslexic learners.

Childhood-Onset ANCA-Associated Vasculitis: From Genetic Studies to Advances in Pathogenesis, Classification and Novel Therapeutic Approaches.

Childhood-onset antineutrophil cytoplasmic antibodies (ANCA)-associated vasculitis (AAV) represents a heterogeneous group of multi-system autoimmune conditions associated with chronic inflammation, characteristically affecting small blood vessels, leading to various organ and system manifestations. Although rare in paediatric populations, AAV poses challenges in early recognition, diagnosis and management of refractory cases. This review highlights the characteristics of clinical presentation and outcomes of AAV in children, as well as its current classification and progress achieved in understanding the disease pathogenesis, with a focus on adult and paediatric genetic studies. Furthermore, we discuss the management of AAV in children, as well as new emerging therapies and future research needs, while proposing a potential algorithm for a childhood-onset-AAV therapeutic approach based on the disease phenotype.

TimbreSense: Timbre Abnormality Detection for Bel Canto with Smart Devices

With the rise of mobile devices, bel canto practitioners increasingly utilize smart devices as auxiliary tools for improving their singing skills. However, they frequently encounter timbre abnormalities during practice, which, if left unaddressed, can potentially harm their vocal organs. Existing singing assessment systems primarily focus on pitch and melody and lack real-time detection of bel canto timbre abnormalities. Moreover, the diverse vocal habits and timbre compositions among individuals present significant challenges in cross-user recognition of such abnormalities. To address these limitations, we propose TimbreSense, a novel bel canto timbre abnormality detection system. TimbreSense enables real-time detection of the five major timbre abnormalities commonly observed in bel canto singing. We introduce an effective feature extraction pipeline that captures the acoustic characteristics of bel canto singing. By applying temporal average pooling to the Short-Time Fourier Transform (STFT) spectrogram, we reduce redundancy while preserving essential frequency-domain information. Our system leverages a transformer model with self-attention mechanisms to extract correlation and semantic features of overtones in the frequency domain. Additionally, we employ a few-shot learning approach involving pre-training, meta-learning, and fine-tuning to enhance the system’s cross-domain recognition performance while minimizing user usage costs. Experimental results demonstrate the system’s strong cross-user domain recognition performance and real-time capabilities.

An Action Evaluation Method for Virtual Reality Simulation Power Training Based on an Improved Dynamic Time Warping Algorithm

To address the shortcomings in action evaluation within VR simulation power training, this paper introduces a novel action recognition and evaluation method based on dynamic recognition of finger keypoints combined with an improved Dynamic Time Warping (DTW) algorithm. By constructing an action recognition model centered on hand keypoints, the proposed method integrates distance similarity and cosine similarity to account comprehensively for both numerical differences and directional consistency of action features. This approach effectively tackles the challenges of feature extraction and recognition for complex actions in VR power training. Furthermore, a scoring mechanism based on the improved DTW algorithm is proposed, which employs Gaussian-weighted feature-derivative Euclidean distance combined with cosine similarity. This method significantly reduces computational complexity while improving scoring accuracy and consistency. Experimental results indicated that the improved DTW algorithm outperformed traditional methods in terms of classification accuracy and robustness. In particular, cosine similarity demonstrated superior performance in capturing dynamic variations and assessing the consistency of fine hand movements. This study provides an essential technical reference for action evaluation in VR simulation power training and offers a scientific basis for advancing the intelligence and digitalization of power VR training environments.

Software Optimization and Design Methodology for Low Power Computer Vision Systems

This tutorial article addresses a low power computer vision system as an example of a growing application domain of neural networks, exploring various technologies developed to enhance accuracy within the resource and performance constraints imposed by the hardware platform. Focused on a given hardware platform and network model, software optimization techniques, including pruning, quantization, low-rank approximation, and parallelization, aim to satisfy resource and performance constraints while minimizing accuracy loss. Due to the interdependence of model compression approaches, their systematic application is crucial, as evidenced by winning solutions in the Lower Power Image Recognition Challenge (LPIRC) of 2017 and 2018. Recognizing the typical heterogeneity of processing elements in contemporary hardware platforms, the effective utilization through parallelizing neural networks emerges as increasingly vital for performance enhancement. The article advocates for a more impactful strategy—designing a network architecture tailored to a specific hardware platform. For detailed information on each technique, the article provides corresponding references.

Poly(p-Phenyleneethynylene)s-Based Sensor Array for Diagnosis of Clinical Diseases.

Inspired by the mammalian taste and olfactory systems, array-based pattern recognition technology has demonstrated significant potential in discerning subtle differences between highly similar compounds and complex mixtures, owing to their unique parallel detection mechanism based on cross-reactive signals. While optical sensor array has been extensively employed in the field of chemical sensing, they encounter significant challenges in non-specific recognition of multiple analytes at low concentrations, particularly in rife environments with complex interferences. Poly(p-phenylene ethynylene)s (PPEs) offer substantial advantages in detecting multi-analytes at low concentrations, owing to its distinctive optical properties, including the "molecular wire" effect, fluorescence super-amplification and super-quenching. This is particularly promising for the parallel detection of ultra-low concentration multi-biomarkers in clinical diseases. As the continuous development of PPEs sensor array, more sensitive methods for rapid detection of clinical disease will be further developed. It will promote the further development of the field of early diagnosis of clinical diseases.

A temporal-spectral graph convolutional neural network model for EEG emotion recognition within and across subjects

EEG-based emotion recognition uses high-level information from neural activities to predict emotional responses in subjects. However, this information is sparsely distributed in frequency, time, and spatial domains and varied across subjects. To address these challenges in emotion recognition, we propose a novel neural network model named Temporal-Spectral Graph Convolutional Network (TSGCN). To capture high-level information distributed in time, spatial, and frequency domains, TSGCN considers both neural oscillation changes in different time windows and topological structures between different brain regions. Specifically, a Minimum Category Confusion (MCC) loss is used in TSGCN to reduce the inconsistencies between subjective ratings and predefined labels. In addition, to improve the generalization of TSGCN on cross-subject variation, we propose Deep and Shallow feature Dynamic Adversarial Learning (DSDAL) to calculate the distance between the source domain and the target domain. Extensive experiments were conducted on public datasets to demonstrate that TSGCN outperforms state-of-the-art methods in EEG-based emotion recognition. Ablation studies show that the mixed neural networks and our proposed methods in TSGCN significantly contribute to its high performance and robustness. Detailed investigations further provide the effectiveness of TSGCN in addressing the challenges in emotion recognition.

Automatic classification of HEp-2 specimens by explainable deep learning and Jensen-Shannon reliability index

The Anti-Nuclear Antibodies (ANA) test using Human Epithelial type 2 (HEp-2) cells in the Indirect Immuno-Fluorescence (IIF) assay protocol is considered the gold standard for detecting Connective Tissue Diseases. Computer-assisted systems for HEp-2 image analysis represent a growing field that harnesses the potential offered by novel machine learning techniques to address the classification of HEp-2 images and ANA patterns.In this study, we introduce an innovative platform based on transfer learning with pre-trained deep learning models. This platform combines the power of unsupervised deep description of HEp-2 images, a novel feature selection approach designed for unbalanced datasets, and independent testing using two distinct datasets from different hospitals to tackle cross-hardware compatibility issues. To enhance the trustworthiness of our method, we also present a modified version of gradient-weighted class activation mapping for regional explainability and introduce a new sample quality index based on the Jensen-Shannon divergence to enhance method reliability and quantify sample heterogeneity.The results we provide demonstrate exceptionally high performance in intensity and ANA pattern recognition when compared to state-of-the-art approaches. Our method's ability to eliminate the need for cell segmentation in favor of statistical analysis of the sample makes it applicable, robust, and versatile. Our future work will focus on addressing the challenge of mitotic spindle recognition by expanding our proposed approach to cover mixed patterns.

Stimuli-Responsive Photoluminescent Molecular Tweezers for Highly Enantioselective Discrimination of Chiral Primary Amines.

To address the challenge of chiral recognition in terms of efficiency and generality, we propose a novel fluorescence sensing approach by rationally designing metal-ion-responsive chiral molecular tweezers. The flexible and adaptable molecular tweezers enable facile recognition of 31 structurally varied chiral primary amine compounds, including amino acids, amino acid esters, and chiral amines. Notably, upon stimulation by zinc ions, the chiral molecular tweezers demonstrate a higher enantioselective fluorescence response. Combined density functional theory calculations reveal that the chiral sensing mechanism relies on differential reaction rates and potential hydrogen-bonding interactions between the two enantiomers and the chiral receptor, which results in one of the enantiomers forming a more abundant, stable, and structurally rigid complex with the receptor, resulting in a significant increase in the fluorescence intensity and enantioselectivity. The stimuli-responsive molecular tweezers approach provides a novel strategy for precise stereocontrol and universality of chiral recognition, offering a promising tool for applications in various fields.