Search Results (1,512)

The exponential growth of user-generated video content necessitates efficient summarization systems for improved accessibility, retrieval, and analysis. This study presents and benchmarks a multimodal video summarization framework that classifies segments as informative or non-informative using audio, visual, and fused features. Sixty hours of annotated video across ten diverse categories were analyzed. Audio features were extracted with pyAudioAnalysis, while visual features (colour histograms, optical flow, object detection, facial recognition) were derived using OpenCV. Six supervised classifiers—Naive Bayes, K-Nearest Neighbors, Logistic Regression, Decision Tree, Random Forest, and XGBoost—were evaluated, with hyperparameters optimized via grid search. Temporal coherence was enhanced using median filtering. Random Forest achieved the best performance, with 74% AUC on fused features and a 3% F1-score gain after post-processing. Spectral flux, grayscale histograms, and optical flow emerged as key discriminative features. The best model was deployed as a practical web service using TensorFlow and Flask, integrating informative segment detection with subtitle generation via beam search to ensure coherence and coverage. System-level evaluation demonstrated low latency and efficient resource utilization under load. Overall, the results confirm the strength of multimodal fusion and ensemble learning for video summarization and highlight their potential for real-world applications in surveillance, digital archiving, and online education. Full article

Maintaining the freshness of shrimp is a critical issue in quality and safety control within the food processing industry. Traditional methods often rely on destructive techniques, which are difficult to apply in online real-time monitoring. To address this challenge, this study aims to propose a non-destructive approach for shrimp freshness assessment based on imaging and deep learning, enabling efficient and reliable freshness classification. The core innovation of the method lies in constructing an improved GoogLeNet architecture. By incorporating the ELU activation function, L2 regularization, and the RMSProp optimizer, combined with a transfer learning strategy, the model effectively enhances generalization capability and stability under limited sample conditions. Evaluated on a shrimp image dataset rigorously annotated based on TVB-N reference values, the proposed model achieved an accuracy of 93% with a test loss of only 0.2. Ablation studies further confirmed the contribution of architectural and training strategy modifications to performance improvement. The results demonstrate that the method enables rapid, non-contact freshness discrimination, making it suitable for real-time sorting and quality monitoring in shrimp processing lines, and providing a feasible pathway for deployment on edge computing devices. This study offers a practical solution for intelligent non-destructive detection in aquatic products, with strong potential for engineering applications. Full article

Automatic crater detection is crucial for planetary science, but still faces several long-standing challenges. The morphological characteristics of craters exhibit significant variability; combined with complex lighting conditions, this makes feature extraction difficult, especially for small or severely degraded features. These difficulties are further compounded by incomplete ground truth annotations, which limit the effectiveness of supervised learning. In addition, achieving a balance between detection accuracy and computational efficiency remains a critical bottleneck, especially in large-scale planetary surveys. Traditional postprocessing algorithms also often struggle to resolve complex spatial hierarchies in densely cratered regions, leading to substantial omissions and misclassifications. To address these interrelated challenges, we propose Crater-MASN, a lightweight adaptive detection framework specifically designed for lunar crater analysis. The architecture employs a compact GhostNet backbone to balance efficiency and accuracy, while enhancing multi-scale feature representation through a novel bidirectional integration and fusion module (BIFM) to better capture the morphological diversity of craters. To mitigate the impact of incomplete annotations, we introduce an adaptive semantic contrastive sampling (ASCS) mechanism which dynamically mines unlabeled craters through semantic clustering and contrastive learning. Additionally, we design the hierarchical soft NMS (H-SoftNMS) algorithm, a geometry-aware postprocessing method that selectively suppresses non-hierarchical overlaps to preserve nested craters, thereby achieving more accurate crater retention in dense regions. Experiments on a dedicated lunar crater dataset demonstrate the effectiveness of Crater-MASN. The model achieves an mAP50 of 91.0% with only 2.1 million parameters. When combined with H-SoftNMS, it achieves a recall rate of 95.0% and new discovery rate ${\mathcal{P}}_{\mathrm{NDR}}$ of 89.6%. These results highlight the potential of Crater-MASN as a scalable and reliable tool for high-precision crater cataloging and planetary surface analysis. Full article

Streptococcus pharyngitis (strep throat) poses a significant health challenge in rural and remote Indigenous communities in Australia, where access to medical resources is limited. Delays in diagnosis and treatment increase the risk of serious complications, including acute rheumatic fever and rheumatic heart disease. This paper presents a proof-of-concept AI-based diagnostic model designed to support clinicians in underserved communities. The model combines a lightweight Swin Transformer–based image classifier with a BLIP-2-based explainable image annotation system. The classifier predicts strep throat from throat images, while the explainable model enhances transparency by identifying key clinical features such as tonsillar swelling, erythema, and exudate, with synthetic labels generated using GPT-4o-mini. The classifier achieves 97.1% accuracy and an ROC-AUC of 0.993, with an inference time of 13.8 ms and a model size of 28 million parameters; these results demonstrate suitability for deployment in resource-constrained settings. As a proof-of-concept, this work illustrates the potential of AI-assisted diagnostics to improve healthcare access and could benefit similar research efforts that support clinical decision-making in remote and underserved regions. Full article

Scientific creativity is a crucial indicator of adolescents’ potential in science and technology, and its automated evaluation plays a vital role in the early identification of innovative talent. To address challenges such as limited sample sizes, high annotation costs, and modality heterogeneity, this study proposes a multimodal assessment method that integrates data augmentation and pseudo-labeling techniques. For the first time, a joint enhancement approach is introduced that combines textual and visual data with a pseudo-labeling strategy to accommodate the characteristics of text–image integration in elementary students’ cognitive expressions. Specifically, SMOTE is employed to expand questionnaire data, EDA is used to enhance hand-drawn text–image data, and text–image semantic alignment is applied to improve sample quality. Additionally, a confidence-driven pseudo-labeling mechanism is incorporated to optimize the use of unlabeled data. Finally, multiple machine learning models are integrated to predict scientific creativity. The results demonstrate the following: 1. Data augmentation significantly increases sample diversity, and the highest accuracy of information alignment was achieved when text and images were matched. 2. The combination of data augmentation and pseudo-labeling mechanisms improves model robustness and generalization. 3. Family environment, parental education, and curiosity are key factors influencing scientific creativity. This study offers a cost-effective and efficient approach for assessing scientific creativity in elementary students and provides practical guidance for fostering their innovative potential. Full article

Background and Objectives: Barrett’s Esophagus (BE) is a precursor to esophageal adenocarcinoma, and early detection is essential to reduce cancer risk. This study aims to develop a YOLO-based ensemble framework to improve the automated detection of BE-associated mucosal lesions on endoscopic images. Methods: A dataset of 3620 annotated endoscopic images was collected from 132 patients. Five YOLO variants, YOLOv5, YOLOv9, YOLOv10, YOLOv11, and YOLOv12, were selected based on their architectural diversity and detection capabilities. Each model was trained individually, and their outputs were integrated using a Non-Maximum Suppression (NMS)-based ensemble strategy. Multiple ensemble configurations were evaluated to assess the impact of fusion depth on detection performance. Results: The ensemble models consistently outperformed individual YOLO variants in recall, the primary evaluation metric. The entire five-model ensemble achieved the highest recall (0.974), significantly reducing missed lesion detections. Statistical analysis using McNemar’s test and bootstrap confidence intervals confirmed the superiority in most comparisons. Conclusions: The proposed YOLO ensemble framework demonstrates enhanced sensitivity and robustness in detecting BE lesions. Its integration into clinical workflows can improve early diagnosis and reduce diagnostic workload, offering a promising tool for computer-aided screening in gastroenterology. Full article

In today’s fast-paced release cycles, mobile app user reviews offer a valuable source for tracking the evolution of user needs. At the core of these needs lies a structure of interdependencies—some enhancements are only relevant in specific usage contexts, while others may conflict when implemented together. Identifying these relationships is essential for anticipating feature interactions, resolving contradictions, and enabling context-aware, user-driven planning. The present work introduces an ontology-enhanced AI framework for predicting whether the requirements mentioned in reviews are interdependent. The core component is a Bidirectional Encoder Representations from Transformers (BERT) classifier retrained within a large-language-model-driven active learning loop that focuses on instances with uncertainty. The framework integrates contextual and structural reasoning; contextual analysis captures the semantic intent and functional role of each requirement, enriching the understanding of user expectations. Structural reasoning relies on a domain-specific ontology that serves as both a knowledge base and an inference layer, guiding the grouping of requirements. The model achieved strong performance on annotated banking app reviews, with a validation F1-score of 0.9565 and an area under the ROC curve (AUC) exceeding 0.97. The study results contribute to supporting developers in prioritizing features based on dependencies and delivering more coherent, conflict-free releases. Full article

The sit-and-reach test is a common stretching exercise suitable for adolescents, aimed at improving joint flexibility and somatic neural control, and has become a mandatory item in China’s student physical fitness assessments. However, many students tend to perform incorrect postures during their practice, which may lead to sports injuries such as muscle strains if sustained over time. To address this issue, this paper proposes a Ghost-ST-GCN model for judging the correctness of the sit-and-reach pose. The model first requires detecting seven body keypoints. Leveraging a publicly available labeled keypoint dataset and unlabeled sit-and-reach videos, these keypoints are acquired through the proposed self-train method using the BlazePose network. Subsequently, the keypoints are fed into the Ghost-ST-GCN model, which consists of nine stacked GCN-TCN blocks. Critically, each GCN-TCN layer is embedded with a ghost layer to enhance efficiency. Finally, a classification layer determines the movement’s correctness. Experimental results demonstrate that the self-train method significantly improves the annotation accuracy of the seven keypoints; the integration of ghost layers streamlines the overall detection model; and the system achieves an action detection accuracy of 85.20% for the sit-and-reach exercise, with a response latency of less than 1 s. This approach is highly suitable for guiding adolescents to standardize their movements during independent sit-and-reach practice. Full article

Understanding discourse context is fundamental to human language comprehension. Despite the remarkable progress achieved by Large Language Models, they still struggle with discourse-level anaphora resolution, particularly in Chinese. One major challenge is zero anaphora, a prevalent linguistic phenomenon in which referential elements are omitted, increasing complexity and ambiguity for computational models. To address this issue, we introduce CDAMR (Chinese Discourse Abstract Meaning Representation), a novel annotated corpus that systematically labels zero pronouns across diverse syntactic positions along with their discourse-level coreference chains. In addition, we present a hierarchical discourse-semantic enhanced model that separately encodes local discourse semantics and global discourse semantics, and models their interactions via structured multi-attention mechanisms. Experiments on both CDAMR and OntoNotes demonstrate the approach’s cross-corpus generalizability and effectiveness, achieving F1 scores of 59.86% and 60.54%, respectively. Ablation studies further confirm that discourse-level semantics significantly enhance zero pronoun resolution. These findings highlight the value of cognitively inspired discourse modeling and the importance of comprehensive discourse annotations for languages with limited explicit syntactic cues such as Chinese. Full article

Tree peony (Paeonia ostii) is widely cultivated in China as a traditional medicine and a new high-quality woody oil crop. Enhancing seed yield has become a primary breeding objective in the industrial development of oil tree peonies. Pollination and successful fertilization are essential for optimal seed yield. However, the molecular mechanisms underlying pollination and fertilization in P. ostii remain unclear. In this study, comparative transcriptomic and genetic analyses were conducted to investigate the pistils under different pollination periods of P. ostii ‘Fengdanbai’. Compared with pre-pollination, differentially expressed genes (DEGs) were screened from pistils 48 h after pollination, when most of the pollen tubes had reached the bottom of the style. Functional annotation indicated that these DEGs were involved in hormone signaling and carbohydrate metabolism pathways. Transcription factors and receptor-like kinases play a key role in pollen development, pollen tube growth, and carpel development. Key DEGs (PoUNE10 and PoLIM1) influenced pollination and fertilization and were characterized. Phylogenetic, promoter, and co-expression analyses suggest that they may affect plant pollination, fertilization, and seed yield through pathways such as hormone signaling and photosynthesis in P. ostii ‘Fengdanbai’. Our findings illustrate the molecular changes after pollination and fertilization in P. ostii ‘Fengdanbai’ and provide the molecular characterization of two key genes. These results provide insights into the molecular mechanisms underlying pollination and fertilization in tree peony and suggest potential candidate genes for molecular breeding aimed at improving seed yield in the species. Full article

The feeding behavior of dairy cows constitutes a complex temporal sequence comprising actions such as head lowering, sniffing, arching, eating, head raising, and chewing. Its precise recognition is crucial for refined livestock management. While existing 2D convolution-based models effectively extract features from individual frames, they lack temporal modeling capabilities. Conversely, due to their high computational complexity, 3D convolutional networks suffer from significantly limited recognition accuracy in high-density feeding scenarios. To address this, this paper proposes a Spatio-Temporal Fusion Network (DAS-Net): it designs a collaborative architecture featuring a 2D branch with a triple-attention module to enhance spatial key feature extraction, constructs a 3D branch based on multi-branch dilated convolution and integrates a 3D multi-scale attention mechanism to achieve efficient long-term temporal modeling. On our Spatio-Temporal Dairy Feeding Dataset (STDF Dataset), which contains 403 video clips and 10,478 annotated frames across seven behavior categories, the model achieves an average recognition accuracy of 56.83% for all action types. This result marks a significant improvement of 3.61 percentage points over the original model. Among them, the recognition accuracy of the eating action has been increased to 94.78%. This method provides a new idea for recognizing dairy cow feeding behavior and can provide technical support for developing intelligent feeding systems in real dairy farms. Full article

Colorectal cancer is a significant health concern that can often be prevented through early detection of precancerous polyps during routine screenings. Although artificial intelligence (AI) methods have shown potential in reducing polyp miss rates, clinical adoption remains limited due to concerns over patient privacy, limited access to annotated data, and the high cost of expert labeling. To address these challenges, we propose an unsupervised domain adaptation (UDA) approach that leverages a fully synthetic colonoscopy dataset, SynthColon, and adapts it to real-world, unlabeled data. Our method builds on the DAFormer framework and integrates a Transformer-based hierarchical encoder, a context-aware feature fusion decoder, and a self-training strategy. We evaluate our approach on the Kvasir-SEG and CVC-ClinicDB datasets. Results show that our method achieves improved segmentation performance of 69% mIoU compared to the baseline approach from the original SynthColon study and remains competitive with models trained on enhanced versions of the dataset. Full article

Background: Developmental dysplasia of the hip (DDH), if not treated, can lead to osteoarthritis and disability. Ultrasound (US) is a primary screening method for the detection of DDH, but its interpretation remains highly operator-dependent. We propose a supervised machine learning (ML) image segmentation model for the automated recognition of anatomical structures in hip US images. Methods: We conducted a retrospective observational analysis based on a dataset of 10,767 hip US images from 311 patients. All images were annotated for eight key structures according to the Graf method and split into training (75.0%), validation (9.5%), and test (15.5%) sets. Model performance was assessed using the Intersection over Union (IoU) and Dice Similarity Coefficient (DSC). Results: The best-performing model was based on the SegNeXt architecture with an MSCAN_L backbone. The model achieved high segmentation accuracy (IoU; DSC) for chondro-osseous border (0.632; 0.774), femoral head (0.916; 0.956), labrum (0.625; 0.769), cartilaginous (0.672; 0.804), and bony roof (0.725; 0.841). The average Euclidean distance for point-based landmarks (bony rim and lower limb) was 4.8 and 4.5 pixels, respectively, and the baseline deflection angle was 1.7 degrees. Conclusions: This ML-based approach demonstrates promising accuracy and may enhance the reliability and accessibility of US-based DDH screening. Future applications could integrate real-time angle measurement and automated classification to support clinical decision-making. Full article

Multimodal Sentiment Analysis (MSA) aims to predict subjective human emotions by leveraging multimodal information. However, existing research inadequately utilizes explicit sentiment semantic information at the lexical level in text and overlooks noise interference from non-dominant modalities, such as irrelevant movements in visual modalities and background noise in audio modalities. To address this issue, we propose a multimodal sentiment analysis model based on knowledge enhancement and text-guided learning (MMKT). The model constructs a sentiment knowledge graph for the textual modality using the SenticNet knowledge base. This graph directly annotates word-level sentiment polarity, strengthening the model’s understanding of emotional vocabulary. Furthermore, global sentiment knowledge features are generated through graph embedding computations to enhance the multimodal fusion process. Simultaneously, a dynamic text-guided learning approach is introduced, which dynamically leverages multi-scale textual features to actively suppress redundant or conflicting information in visual and audio modalities, thereby generating purer cross-modal representations. Finally, concatenated textual features, cross-modal features, and knowledge features are utilized for sentiment prediction. Experimental results on the CMU-MOSEI and Twitter2019 dataset demonstrate the superior performance of the MMKT model. Full article

Cotton fiber quality improvement remains a fundamental challenge in breeding programs due to the complex genetic architecture underlying fiber development. The narrow genetic base of upland cotton (Gossypium hirsutum L.) and the quantitative nature of fiber quality traits necessitate innovative approaches for identifying and incorporating superior alleles from related species. We developed a BC₆F₂ population by introgressing chromosome segments from the sea island cotton variety Xinhai 36 (G. barbadense) into the upland cotton variety Xinluzhong 60 (G. hirsutum). Based on fiber strength phenotyping, we constructed two DNA bulks representing extreme phenotypes (20 superior and 12 inferior individuals) for bulked segregant analysis sequencing (BSA-Seq). High-throughput sequencing generated 225.13 Gb of raw data with average depths of 20× for parents and 30× for bulks. SNP calling and annotation were performed using GATK and ANNOVAR against the upland cotton reference genome (TM-1). BSA-Seq analysis identified 13 QTLs primarily clustered within a 1.6 Mb region (20.6–22.2 Mb) on chromosome A10. Within this region, we detected nonsynonymous mutation genes involving a total of six genes. GO and KEGG enrichment analyses revealed significant enrichment for carbohydrate metabolic processes, protein modification, and secondary metabolite biosynthesis pathways. Integration with transcriptome data prioritized GH_A10G1043, encoding a β-amylase family protein, as the key candidate gene. Functional validation through overexpression and RNAi knockdown in Arabidopsis thaliana demonstrated that GH_A10G1043 significantly regulates starch content and β-amylase activity, though without visible morphological alterations. This study successfully identified potential genomic regions and candidate genes associated with cotton fiber strength using chromosome segment substitution lines combined with BSA-Seq. The key candidate gene GH_A10G1043 provides a valuable target for marker-assisted selection in cotton breeding programs. Our findings establish a foundation for understanding the molecular mechanisms of fiber quality formation and offer genetic resources for developing superior cotton varieties with enhanced fiber strength. Full article