Contextual Features Research Articles

Chinese word segmentation with heterogeneous graph convolutional network

Abstract Recently, deep learning methods have achieved remarkable success in the Chinese word segmentation (CWS) task. Some of them enhance the CWS model by utilizing contextual features and external resources (e.g., sub-words, lexicon, and syntax). However, existing approaches fail to fully use the heterogeneous features and their structural information. Therefore, in this paper, we propose a heterogeneous information learning framework for CWS, named heterogeneous graph neural segmenter (HGNSeg), which exploits heterogeneous features with the graph convolutional networks and the pretrained language model. Experimental results on six benchmark datasets (e.g., SIGHAN 2005 and SIGHAN 2008) confirm that HGNSeg can effectively improve the performance of CWS. Importantly, HGNSeg also demonstrates an excellent ability to alleviate the out-of-vocabulary (OOV) issue in cross-domain scenarios.

Well-being and community resilience: The relation between individual, collective, and structural features in an intractable conflict context.

Well-being and community resilience: The relation between individual, collective, and structural features in an intractable conflict context.

FG-UNet: fine-grained feature-guided UNet for segmentation of weeds and crops in UAV images.

Semantic segmentation of weed and crop images is a key component and prerequisite for automated weed management. For weeds in unmanned aerial vehicle (UAV) images, which are usually characterized by small size and easily confused with crops at early growth stages, existing semantic segmentation models have difficulties to extract sufficiently fine features. This leads to their limited performance in weed and crop segmentation of UAV images. We proposed a fine-grained feature-guided UNet, named FG-UNet, for weed and crop segmentation in UAV images. Specifically, there are two branches in FG-UNet, namely the fine-grained feature branch and the UNet branch. In the fine-grained feature branch, a fine feature-aware (FFA) module was designed to mine fine features in order to enhance the model's ability to segment small objects. In the UNet branch, we used an encoder-decoder structure to realize high-level semantic feature extraction in images. In addition, a contextual feature fusion (CFF) module was designed for the fusion of the fine features and high-level semantic features, thus enhancing the feature discrimination capability of the model. The experimental results showed that our proposed FG-UNet, achieved state-of-the-art performance compared to other semantic segmentation models, with mean intersection over union (MIOU) and mean pixel accuracy (MPA) of 88.06% and 92.37%, respectively. The proposed method in this study lays a solid foundation for accurate detection and intelligent management of weeds. It will have a positive impact on the development of smart agriculture. © 2024 Society of Chemical Industry.

A text sentiment analysis method based on BiGRU and capsule network

Most of the existing sentiment analysis methods based on product review texts rarely consider the aspect features of the review texts, and the relevant analysis models do not consider the long-term contextual dependency features and local text features at the same time, thus affecting the accuracy of sentiment analysis. A text sentiment analysis method based on a bidirectional gated recurrent network (BiGRU) and a capsule network is proposed. This method first uses a word frequency statistics-based method to extract the aspect features of the review text and integrates them into the word vector representation, thereby effectively improving the expressive power of the word vector. Then, BiGRU is used to extract the long-term contextual dependency features of the text, and the capsule network is used to extract the local features of the text, thereby achieving high-precision text sentiment analysis based on aspects. Experimental results on real datasets show that the proposed method is superior to existing sentiment analysis models such as bidirectional long short-term memory network (BiLSTM), CNN-LSTM, and TextCNN in terms of evaluation indicators such as accuracy, precision, recall, and score.

Domain knowledge-powered attention for air traffic management hazardous events classification

The automated classification of hazardous events in Air Traffic Management (ATM) involves concise texts with specialized terminology. The problem is currently addressed by machine learning text classification methods; however, some methods overlook domain knowledge, while others require it excessively. This research aims to identify a process for integrating low-cost, text-based domain knowledge into classification to address the issue of domain shift. To achieve this, the Wide and Deep Bidirectional Encoder Representations from Transformers (WD-BERT) model is proposed, featuring a unique knowledge-powered attention mechanism. WD-BERT's Wide Attention module assesses the compatibility between input text and domain knowledge to determine class probabilities, while its Deep Attention module extracts contextual features guided by domain knowledge. Additionally, a text mining method is employed to extract domain knowledge from ATM regulation texts. The model is trained and evaluated using a dataset derived from the China ATM Hazard Source Database. It achieves a multi-label classification accuracy of 80.24% and an F1-micro score of 0.88, outperforming comparative models with state-of-the-art performance. Breakdown analyses of the model indicate that the carefully designed attention mechanism endows WD-BERT with the ability to integrate domain knowledge, allowing it to excel in contexts involving complex sentences, multiple terms, and domain shifts. The innovation of this study lies in the proposed domain knowledge-powered attention mechanism, which allows simply organized domain knowledge to effectively guide ATM hazardous events classification, alleviating the impact of domain shift. This method can also be applied to other vertical domains without knowledge graphs.

TRRHA: A two-stream re-parameterized refocusing hybrid attention network for synthesized view quality enhancement

In multi-view video systems, the decoded texture video and its corresponding depth video are utilized to synthesize virtual views from different perspectives using the depth-image-based rendering (DIBR) technology in 3D-high efficiency video coding (3D-HEVC). However, the distortion of the compressed multi-view video and the disocclusion problem in DIBR can easily cause obvious holes and cracks in the synthesized views, degrading the visual quality of the synthesized views. To address this problem, a novel two-stream re-parameterized refocusing hybrid attention (TRRHA) network is proposed to significantly improve the quality of synthesized views. Firstly, a global multi-scale residual information stream is applied to extract the global context information by using refocusing attention module (RAM), and the RAM can detect the contextual feature and adaptively learn channel and spatial attention feature to selectively focus on different areas. Secondly, a local feature pyramid attention information stream is used to fully capture complex local texture details by using re-parameterized refocusing attention module (RRAM). The RRAM can effectively capture multi-scale texture details with different receptive fields, and adaptively adjust channel and spatial weights to adapt to information transformation at different sizes and levels. Finally, an efficient feature fusion module is proposed to effectively fuse the extracted global and local information streams. Extensive experimental results show that the proposed TRRHA achieves significantly better performance than the state-of-the-art methods. The source code will be available at https://github.com/647-bei/TRRHA.

Multi-Scale Feature Attention Fusion for Image Splicing Forgery Detection

Image splicing is a widely occurrence image tampering technology. With the rapid development of digital image processing technology, detecting image splicing forgery has become significantly challenging. Although various methods have been devised to identify such tampered images, existing approaches have not achieved optimal performance due to limitations in effectively leveraging feature maps of different scales. To address this issue, we propose a novel method for image splicing forgery detection called multi-scale feature attention fusion network (MFAF-Net). We propose a multi-scale atrous feature attention (MAFA) module designed to capture rich contextual features for multi-scale high-level feature fusion. Additionally, we present the multi-branch attention mechanism (MBAM) module to fuse contextual information from various branches for low-level features. This integration enhances the capability of low-level features to produce more refined pixel-level attention. We employ the weighted binary cross-entropy loss and dice loss in the MFAF-Net to overcome the imbalance between positive and negative samples. Extensive experiments demonstrate that the proposed MFAF-Net outperforms state-of-the-art methods. Robustness experiments also show our model exhibits image splicing forgery detection robustness under common attacks.

Performance assessment of food safety systems (FSMSs and FSC) in water bottling factories in view of their contextual characteristics

Foodborne illnesses affect 30% of people in developed nations and lead to financial difficulties in less wealthy countries. Insufficient understanding of pathogens and inadequate food safety practices in water bottling facilities impede progress. To prevent illnesses, ensuring clean water, proper sanitation, and good hand hygiene are essential in order to decrease the 829,000 annual deaths related to diarrhea. This study aimed to evaluate the performance of food safety systems in water bottling factories in view of their contextual characteristics. A study was conducted in northwestern Ethiopia at water bottling factories from January 2023 to August 2023. This study used a mixed method approach to collect and analyze data through laboratory testing, observations, interviews, questionnaires, and document analysis. The Kruskal‒Wallis H test was used to compare mean scores of context riskiness and food safety output across four clusters, while the Mann‒Whitney U test was used to compare means between certified and noncertified factories. In our study, 43% of the samples showed positive total plate count. The result also indicated the presence of potentially dangerous by-products of chlorine, trihalomethanes was found in bottled water. On average, the manufacturers operated in a moderate risk context. Inadequate food safety outcomes resulted from lower food safety management systems scores. Factories with insufficient Food Safety Management Systems often face challenges marked by safety protocol discrepancies, leading to higher risks of contamination and non-compliance with regulations. This assessment underscores the need for tailored interventions focusing on specific contextual barriers to improve food safety outcomes in the bottled water industry.

Perceived discrimination and refraining from seeking physician’s care in Sweden: an intersectional analysis of individual heterogeneity and discriminatory accuracy (AIHDA)

BackgroundDiscrimination may further impede access to medical care for individuals in socially disadvantaged positions. Sociodemographic information and perceived discrimination intersect and define multiple contexts or strata that condition the risk of refraining from seeking physician’s care. By applying analysis of individual heterogeneity and discriminatory accuracy (AIHDA) we aimed to improve the mapping of risk by considering both strata average risk differences and the accuracy of such strata risks for distinguishing between individuals who did or did not refrain from seeking physician’s care.MethodsWe analysed nine annual National Public Health Surveys (2004, 2007–2014) in Sweden including 73,815 participants. We investigated the risk of refraining from seeking physician’s care across 64 intersectional strata defined by sex, education, age, country of birth, and perceived discrimination. We calculated strata-specific prevalences and prevalence ratios (PR) with 95% confidence intervals (CI), and the area under the receiver operating characteristic curve (AUC) to evaluate the discriminatory accuracy (DA).ResultsDiscriminated foreign-born women aged 35–49 with a low educational level show a six times higher risk (PR = 6.07, 95% CI 5.05–7.30) than non-discriminated native men with a high educational level aged 35–49. However, the DA of the intersectional strata was small (AUC = 0.64). Overall, discrimination increased the absolute risk of refraining from seeking physician’s care, over and above age, sex, and educational level.ConclusionsAIHDA disclosed complex intersectional inequalities in the average risk of refraining from seeking physician’s care. This risk was rather high in some strata, which is relevant from an individual perspective. However, from a population perspective, the low DA of the intersectional strata suggests that potential interventions to reduce such inequalities should be universal but tailored to the specific contextual characteristics of the strata. Discrimination impairs access to healthcare.

Named Entity Recognition for Equipment Fault Diagnosis Based on RoBERTa-wwm-ext and Deep Learning Integration

Equipment fault diagnosis NER is to extract specific entities from Chinese equipment fault diagnosis text, which is the premise of constructing an equipment fault diagnosis knowledge graph. Named entity recognition for equipment fault diagnosis can also provide important data support for equipment maintenance support. Equipment fault diagnosis text has complex semantics, fuzzy entity boundaries, and limited data size. In order to extract entities from the equipment fault diagnosis text, this paper presents an NER model for equipment fault diagnosis based on RoBERTa-wwm-ext and Deep Learning network integration. Firstly, this model uses the RoBERTa-wwm-ext to extract context-sensitive embeddings of text sequences. Secondly, the context feature information is obtained through the BiLSTM network. Thirdly, the CRF is combined to output the label sequence with a constraint relationship, improve the accuracy of sequence labeling task, and complete the entity recognition task. Finally, experiments and predictions are carried out on the constructed dataset. The results show that the model can effectively identify five types of equipment fault diagnosis entities and has higher evaluation indexes than the traditional model. Its precision, recall, and F1 value are 94.57%, 95.39%, and 94.98%, respectively. The case study proves that the model can accurately recognize the entity of the input text.

Nonfatal Firearm Injury & Subsequent Emergency Department Utilization Among Non-Elderly Adults

IntroductionThe mortality, long-term morbidity, and exacerbated healthcare needs due to firearm injury in the US are significant and growing. However, the relationship between exposure to a nonfatal firearm injury and long-term ED utilization is poorly understood. This study estimates the association between exposure to a nonfatal firearm injury and ED utilization in the subsequent year. MethodsUsing all-payer ED data among non-elderly adults in Georgia and New York, all ED visits for nonfatal firearm injuries from 2017-2018 were identified. Sociodemographic, clinical, and contextual characteristics between nonfatal firearm injury ED patients and the broader population of ED users were compared. ED utilization in the year following a nonfatal firearm injury relative to ED use in the year before, and compared to ED use by a propensity score matched control group was examined, using Poisson and negative binomial multivariable regressions. Analyses were performed in 2024. ResultsNonfatal firearm injury ED patients were disproportionately male, younger, non-Hispanic Black, uninsured, and residents of areas with low median income and high firearm ownership. Compared to a matched control group, multivariable analyses indicated that nonfatal firearm injury ED patients had significantly higher risks of having hospital admissions through the ED (aRR: 1.42), all-cause injury-related ED visits (aRR: 1.47), non-firearm injury-related ED visits (aRR: 1.26), and additional nonfatal firearm injury-related ED visits (aRR: 325.45) in the subsequent year (p<0.001 for all). About one in every eight ED users with a firearm-related injury at index also sought ED care for another nonfatal firearm injury within one year. ConclusionsNonfatal firearm-related injuries contribute to preventable harm, health inequity and increased ED utilization.

Using query semantic and feature transfer fusion to enhance cardinality estimating of property graph queries

With the increasing complexity and diversity of query tasks, cardinality estimation has become one of the most challenging problems in query optimization. In this study, we propose an efficient and accurate cardinality estimation method to address the cardinality estimation problem in property graph queries, particularly in response to the current research gap regarding the neglect of contextual semantic features. We first propose formal representations of the property graph query and define its cardinality estimation problem. Then, through the query featurization, we transform the query into a vector representation that can be learned by the estimation model, and enrich the feature vector representation by the context semantic information of the query. We finally propose an estimation model for property graph queries, specifically introducing a feature information transfer module to dynamically control the information flow meanwhile achieving the model’s feature fusion and inference. Experimental results on three datasets show that the estimation model can accurately and efficiently estimate the cardinality of property graph queries, the mean Q_error and RMSE are reduced by about 30% and 25% than the state-of-art estimation models. The context semantics features of queries can improve the model’s estimation accuracy, the mean Q_error result is reduced by about 20% and the RMSE result is about 5%.

Tumor detection in breast cancer pathology patches using a Multi-scale Multi-head Self-attention Ensemble Network on Whole Slide Images

Breast cancer (BC) is the most common type of cancer among women globally and is one of the leading causes of cancer-related deaths among women. In the diagnosis of BC, histopathological assessment is the gold standard, where automated tumor detection technologies play a pivotal role. Utilizing Convolutional Neural Networks (CNNs) for automated analysis of image patches from Whole Slide Images (WSIs) enhances detection accuracy and alleviates the workload of pathologists. However, CNNs often face limitations in handling pathological patches due to a lack of sufficient contextual information and limited feature generation capabilities. To address this, we propose a novel Multi-scale Multi-head Self-attention Ensemble Network (MMSEN), which integrates a multi-scale feature generation module, a convolutional self-attention module, and an adaptive feature integration and output module, effectively optimizing the performance of classical CNNs. The design of MMSEN optimizes the capture of key information and the comprehensive integration of features in WSIs pathological patches, significantly enhancing the precision of tumor detection. Validation results from a five-fold cross-validation experiment on the PatchCamelyon (PCam) dataset demonstrate that MMSEN achieves a ROC-AUC of 99.01%±0.02%, an F1 Score of 98.00%±0.08%, a Balanced Accuracy (B-Acc) of 98.00%±0.08%, and a Matthews Correlation Coefficient (MCC) of 96.00%±0.16% (p<0.05). These results demonstrate the effectiveness and potential of MMSEN in detecting tumors from pathological patches in WSIs for BC.

Anomaly Detection and Approximate Similarity Searches of Transients in Real-time Data Streams

We present Lightcurve Anomaly Identification and Similarity Search (LAISS), an automated pipeline to detect anomalous astrophysical transients in real-time data streams. We deploy our anomaly detection model on the nightly Zwicky Transient Facility (ZTF) Alert Stream via the ANTARES broker, identifying a manageable ∼1–5 candidates per night for expert vetting and coordinating follow-up observations. Our method leverages statistical light-curve and contextual host galaxy features within a random forest classifier, tagging transients of rare classes (spectroscopic anomalies), of uncommon host galaxy environments (contextual anomalies), and of peculiar or interaction-powered phenomena (behavioral anomalies). Moreover, we demonstrate the power of a low-latency (∼ms) approximate similarity search method to find transient analogs with similar light-curve evolution and host galaxy environments. We use analogs for data-driven discovery, characterization, (re)classification, and imputation in retrospective and real-time searches. To date, we have identified ∼50 previously known and previously missed rare transients from real-time and retrospective searches, including but not limited to superluminous supernovae (SLSNe), tidal disruption events, SNe IIn, SNe IIb, SNe I-CSM, SNe Ia-91bg-like, SNe Ib, SNe Ic, SNe Ic-BL, and M31 novae. Lastly, we report the discovery of 325 total transients, all observed between 2018 and 2021 and absent from public catalogs (∼1% of all ZTF Astronomical Transient reports to the Transient Name Server through 2021). These methods enable a systematic approach to finding the “needle in the haystack” in large-volume data streams. Because of its integration with the ANTARES broker, LAISS is built to detect exciting transients in Rubin data.

Robust ELM-PID tracing control on autonomous mobile robot via transformer-based pavement crack segmentation

Pavement crack tracing is paramount to missions encompassing automated crack sealing for road maintenance. However, existing methods still face several challenges, including the incapability to precisely extract crack trajectories and the challenge of tuning control parameters within intricate backgrounds. To address these limitations, the ViT-S2T network and the ELM-PID control system are proposed for crack tracing. Specifically, the ViT-S2T consists of two branches. The transformer-based feature extraction module (TFEM) integrates multi-head attention mechanism and multi-layer perceptron to capture global contextual crack semantic features. The incoherent segmentation masks (ISM) employs a binary classifier to predict the coarsest irrelevant mask and further performs up-sampling fusion of higher-resolution features. Moreover, the Neural-PID control method is designed to track crack trajectories, combining Extreme Learning Machines (ELM) and Proportional Integral Derivative (PID). The ELM-PID controller utilizes a three-layer backpropagation neural network and proposes the ELM model for adaptive adjustment by predicting the tuning parameters of PID. This framework is applied to real-time visual tracing for edge AI. Extensive tests performed on three arduous datasets of DeepCrack, CFD, and S2TCrack, achieving a precision of 82.76% and mAP@0.5 of 75.63% and speed of 0.0513 m/s, demonstrating the superior and robust nature of our approach in pavement crack tracing.

Uncovering the Profile of Ubiquitination Motif in Catalytic Proteins using Sequence Contextual Features

Ubiquitin molecules modify their target protein by conjugating the amino group of lysine side chain, thereby regulating the activity and function of most intracellular proteins. However, within multiple lysine residues on a target protein, only particular residues can be modified by ubiquitin molecules. The occurrence of ubiquitination varies on different lysine residues. The ubiquitination of a lysine residue is ubiquitinated depends on the environment of amino acid sequences near the lysine residue; this sequence is called a motif. Although the interaction of motifs with ubiquitin molecules have been extensively studied, there have been no reports on the ubiquitination motifs. In the current study, 133878 ubiquitinated peptide data from 27762 proteins in 30 articles and the mUbiSiDa database in the past decade were collected and analyzed. A total of 208 motifs were found with a score above 6, covering most of the ubiquitination peptides. Of these, six motifs with higher scores and significant regularity were found to belong to zinc finger proteins and serine/threonine kinases, termed zinc finger and serine/threonine kinase motifs, respectively. Analysis on four zinc finger motifs revealed that the ubiquitination sites in zinc finger motifs, xxHxxxxxxE_K_xxxCxxCxxx and xxxKxxHxxx_K_xxxxxExxKx, were located between two zinc finger domains, while the ubiquitination sites in zinc finger motifs, xxxxxxxxCG_K_xFxxxxxxxx and CxxxxxxFxQ_K_xxxxxxxxxx were located within zinc chelation structure near one Cys residue. Different proteins contain different types and quantities of zinc finger motifs. The results of the 3D structure showed that the ubiquitination sites in motif CxxxxxxFxQ_K_xxxxxxxxxx were located in the ring structure of the zinc finger domain. This may affect the interaction between zinc finger domains and DNA. Meanwhile, the ubiquitination sites of the other three zinc finger motifs located outside the interaction structure between zinc finger domains and DNA may not serve to block interaction but degradation. The analysis of two serine/threonine kinase motifs revealed that some residues near the ubiquitination site in two serine/threonine kinase motifs were completely identical in different proteins; however, these conserved residues were completely different in different motifs, suggesting the varied function of the two motifs. Moreover, it was found that no protein simultaneously contains both serine/threonine kinase motifs. Although there were significant differences in the amino acid sequences of these motifs, their spatial structures were highly conserved. These results suggest that the regulatory functions of the two motifs may differ. This is because two motifs may be recognized by different ubiquitinases or deubiquitinases. Overall, these results suggest that the motifs of ubiquitination sites exhibit high regularity and special structural characteristics in different types of proteins. This study sheds light on the profile and regularity of ubiquitination motifs, promoting the prediction of the function, structure, and modified sites of ubiquitinated proteins.

HOI-V: One-stage human-object interaction detection based on multi-feature fusion in videos

Effective detection of Human-Object Interaction (HOI) is important for machine understanding of real-world scenarios. Nowadays, image-based HOI detection has been abundantly investigated, and recent one-stage methods strike a balance between accuracy and efficiency. However, it is difficult to predict temporal-aware interaction actions from static images since limited temporal context information is introduced. Meanwhile, due to the lack of early large-scale video HOI datasets and the high computational cost of spatial-temporal HOI model training, recent exploratory studies mostly follow a two-stage paradigm, but independent object detection and interaction recognition still suffer from computational redundancy and independent optimization. Therefore, inspired by the one-stage interaction point detection framework, a one-stage spatial-temporal HOI detection baseline is proposed in this paper, in which the short-term local motion features and long-term temporal context features are obtained by the proposed temporal differential excitation module (TDEM) and DLA-TSM backbone. Complementary visual features between multiple clips are then extracted by multi-feature fusion and fed into the parallel detection branches. Finally, a video dataset containing only actions with reduced data size (HOI-V) is constructed to motivate further research on end-to-end video HOI detection. Extensive experiments are also conducted to verify the validity of our proposed baseline.

CoFiNet: Unveiling camouflaged objects with multi-scale finesse

Camouflaged Object Detection (COD) is a critical aspect of computer vision aimed at identifying concealed objects, with applications spanning military, industrial, medical and monitoring domains. To address the problem of poor detail segmentation effect, we introduce a novel method for camouflaged object detection, named CoFiNet. Our approach primarily focuses on multi-scale feature fusion and extraction, with special attention to the model’s segmentation effectiveness for detailed features, enhancing its ability to effectively detect camouflaged objects. CoFiNet adopts a coarse-to-fine strategy. A multi-scale feature integration module is laveraged to enhance the model’s capability of fusing context feature. A multi-activation selective kernel module is leveraged to grant the model the ability to autonomously alter its receptive field, enabling it to selectively choose an appropriate receptive field for camouflaged objects of different sizes. During mask generation, we employ the dual-mask strategy for image segmentation, separating the reconstruction of coarse and fine masks, which significantly enhances the model’s learning capacity for details. Comprehensive experiments were conducted on four different datasets, demonstrating that CoFiNet achieves state-of-the-art performance across all datasets. The experiment results of CoFiNet underscore its effectiveness in camouflaged object detection and highlight its potential in various practical application scenarios.

Psychosocial and Moral Factors of Bystanders in Peer Bullying

The aim of our study was to explore students’ bystander roles in peer bullying considering the interaction between their individual and contextual characteristics. We included social status goals (popularity, social preference and social insecurity), moral disengagement and peer support as key variables for differentiating bystander behaviour. Our sample comprised 5148 students from the 8th and 9th grades across 118 primary schools in Slovenia. Bystander roles (active and passive reinforcers, ignorants and defenders) were determined by peer nominations. Other characteristics were measured with self-report questionnaires. For each variable, we normalized students’ results with regard to their classroom to define classroom norms as the context of peer bullying resulting in “low”, “average” and “high” group. Students with similar individual and contextual characteristics were grouped into four clusters. The results showed that students from all clusters were present in all bystander roles, which indicates a highly complex interaction of various factors in bystander behaviour. Some educational implications are discussed.

S2VSNet: Single stage V-shaped network for image deraining & dehazing

Producing high-quality, noise-free images from noisy or hazy inputs relies on essential tasks such as single image deraining and dehazing. In many advanced multi-stage networks, there is often an imbalance in contextual information, leading to increased complexity. To address these challenges, we propose a simplified method inspired by a U-Net structure, resulting in the “Single-Stage V-Shaped Network” (S2VSNet), capable of handling both deraining and dehazing tasks. A key innovation in our approach is the introduction of a Feature Fusion Module (FFM), which facilitates the sharing of information across multiple scales and hierarchical layers within the encoder-decoder structure. As the network progresses towards deeper layers, the FFM gradually integrates insights from higher levels, ensuring that spatial details are preserved while contextual feature maps are balanced. This integration enhances the image processing capability, producing noise-free, high-quality outputs. To maintain efficiency and reduce system complexity, we replaced or removed several non-essential non-linear activation functions, opting instead for simple multiplication operations. Additionally, we introduced a “Multi-Head Attention Integrated Module” (MHAIM) as an intermediary layer between encoder-decoder levels. This module addresses the limited receptive fields of traditional Convolutional Neural Networks (CNNs), allowing for the capture of more comprehensive feature-map information. Our focus on deraining and dehazing led to extensive experiments on a wide range of synthetic and real-world datasets. To further validate the robustness of our network, we implemented S2VSNet on a low-end edge device, achieving deraining in 2.46 seconds.