Attention Weights Research Articles

Triple-Stream Commonsense Circulation Transformer Network for Image Captioning

Traditional image captioning methods only have a local perspective at the dataset level, allowing them to explore dispersed information within individual images. However, the lack of a global perspective prevents them from capturing common characteristics among similar images. To address the limitation, this paper introduces a novel Triple-stream Commonsense Circulating Transformer Network (TCCTN). It incorporates contextual stream into the encoder, combining enhanced channel stream and spatial stream for comprehensive feature learning. The proposed commonsense-aware contextual attention (CCA) module queries commonsense contextual features from the dataset, obtaining global contextual association information by projecting grid features into the contextual space. The pure semantic channel attention (PSCA) module leverages compressed spatial domain for channel pooling, focusing on attention weights of pure channel features to capture inherent semantic features. The region spatial attention (RSA) module enhances spatial concepts in semantic learning by incorporating region position information. Furthermore, leveraging the complementary differences among the three features, TCCTN introduces the mixture of experts strategy to enhance the unique discriminative ability of features and promote their integration in textual feature learning. Extensive experiments on the MS-COCO dataset demonstrate the effectiveness of contextual commonsense stream and the superior performance of TCCTN.

Lung-YOLO: Multiscale feature fusion attention and cross-layer aggregation for lung nodule detection

ObjectiveLung cancer is a significant public health problem worldwide, and its mortality and morbidity rates are among the highest of cancers. At the same time, early diagnosis of nodules can significantly improve the survival rate of patients. Therefore, this paper proposes the Lung-YOLO algorithm for lung CT image detection based on YOLOv6. MethodsFirst, to enable the network to detect nodules of different sizes and to minimize missed detections, we introduce the Multi-scale Dual-branch Attention (MSDA) mechanism in the feature extraction part of the network. The input features undergo continual dilation convolutions, effectively establish remote dependencies, and the fused multiscale contextual information expands the receptive field while enhancing the model’s ability to detect targets of different sizes, which not only contains precise category and location information but also enables the allocation of attentional weights to generate more pixel-level attention. Then, the fused features access the dual-branch attention module to shift the model’s attention to the target nodules, and the composed dual-branch structure captures cross-dimensional interactions and realizes inter-dimensional dependencies, effectively improving the detection performance. Second, during the feature transfer process, the original Bidirectional Feature Pyramid structure (RepBiFPAN) suffers from the loss of detailed information such as texture and color, making it challenging to localize target nodules accurately. To address this, we propose the Cross-layer Aggregation Module (CLAM), by cross-layer aggregating the multi-level feature layer of the backbone with the multi-level detection layer of the head, which preserves the multi-level fine-grained information that may be lost during the feature transfer process, which is crucial for the detection of small targets. Finally, the module proposed in this paper can be easily incorporated into any detection framework for plug-and-play. ResultsOur method achieves accuracy, precision, recall, and mAP of 97.5 %, 96.5 %, 96.9 %, and 97.9 % on the LUNA16 dataset, and 95.1 %, 94.3 %, 93.4 %, and 95.9 % on the LIDC-IDRI dataset, respectively, surpassing many existing state-of-the-art detection methods. Moreover, the inference speed is 22.8 ms and 28.4 ms per image with 30.6 M parameters, respectively.

High-precision dynamic gesture recognition based on microfiber sensor and EMT-Net

Dynamic gesture recognition, which utilizes flexible wearable sensors and deep learning, is invaluable for human–computer interaction. Nevertheless, the primary challenges persist are the rapid detection of intricate gestures and the accurate recognition of dynamic signals. In this study, we suggest utilizing a microfiber sensor to identify the variations in wrist skin and detect dynamic gesture. In order to tackle the issue of insufficient feature extraction in the detected signals, resulting in reduced accuracy in recognition, we introduce a network dubbed EMT-Net (improve multi-head attention transformer network). This network utilizes a transformer encoder to capture and represent the characteristics of dynamic gesture signal and uses a CNN to classify the encoded features. To ensure that the model comprehensively captures the temporal and statistical characteristics of the signals, we enhance the multi-head attention mechanism by restricting certain attention heads to concentrate solely on the statistical features of the signals while allowing others to focus on the temporal features and global dependencies. Furthermore, because of the varying discriminatory abilities of different characteristics, we have developed an attention module to redistribution the attention weights on statistical features. The experimental results demonstrate that the microfiber sensor effectively recognizes ten distinct forms of dynamic gesture signals. Simultaneously, EMT-Net attains proficient identification with an accuracy of 98.80%, precision of 98.81%, recall of 98.80%, and an F1 score of 98.80%. The application value of this dynamic gesture recognition technology, which utilizes microfiber sensors and EMT-Net, is significant. The forthcoming alterations in human–computer interaction, virtual reality, and various other domains are anticipated.

MOSGAT: Uniting Specificity-Aware GATs and Cross Modal-Attention to Integrate Multi-Omics Data for Disease Diagnosis.

With the advancement of sequencing methodologies, the acquisition of vast amounts of multi-omics data presents a significant opportunity for comprehending the intricate biological mechanisms underlying diseases and achieving precise diagnosis and treatment for complex disorders. However, as diverse omics data are integrated, extracting sample-specific features within each omics modality and exploring potential correlations among different modalities while avoiding mutual interference becomes a critical challenge in multi-omics data integration research. In the context of this study, we proposed a framework that unites specificity-aware GATs and cross-modal attention to integrate different omics data (MOSGAT). To be specific, we devise Graph Attention Networks (GATs) tailored for each omics modality data to perform feature extraction on samples. Additionally, an adaptive confidence attention weighting technique is incorporated to enhance the confidence in the extracted features. Finally, a cross-modal attention mechanism was devised based on multi-head self-attention, thoroughly uncovering potential correlations between different omics data. Extensive experiments were conducted on four publicly available medical datasets, highlighting the superiority of the proposed framework when compared to state-of-the-art methodologies, particularly in the realm of classification tasks. The experimental results underscore MOSGAT's effectiveness in extracting features and exploring potential inter-omics associations.

A spatiotemporal data mining method for advanced prediction and assessment of large combined excavation-induced wall deformations and risks

A holistic and precise assessment of retaining wall deformations is critical for on-site risk management of large combined deep excavation projects, where the risk-related points are highly dispersed, evolving, and interacting. Despite extensive exploration of this topic in previous studies, the omission of intricate spatiotemporal characteristics of wall deformations has resulted in diminished prediction accuracy and stability. To mitigate this deficiency, a spatiotemporal characteristics matrix for all data points and time series was first generated for a deep excavation scenario and used as input for a new hybrid model that combines convolutional neural network (CNN) and long short-term memory (LSTM) with incorporated attention mechanism (CNN-LSTM-Att), which enables the cross-learning mechanism and improves interpretability. In addition, by leveraging the attention weight, a new risk assessment index for retaining wall deformations across various scenarios was formulated. Then the proposed method was applied in a large combined deep excavation in Shanghai, China. The results show that: (1) The incorporation of fed-in characteristic data and the attention mechanism enables the proposed method to produce satisfactory prediction results for the holistic spatiotemporal distribution of a large combined excavation; (2) Compared with other published models, the proposed model shows much better prediction accuracy, interpretability, and stability, especially in medium- and long-term predictions; and (3) The new risk assessment index serves as a reliable decision-making tool for assessing the risk evolution of retaining wall deformations and provides valuable guidance for effective risk management in multi-scenario excavation projects.

PRE-YOLO: A Lightweight Model for Detecting Helmet-Wearing of Electric Vehicle Riders on Complex Traffic Roads

Electric vehicle accidents on the road occur frequently, and head injuries are often the cause of serious casualties. However, most electric vehicle riders seldom wear helmets. Therefore, combining target detection algorithms with road cameras to intelligently monitor helmet-wearing has extremely important research significance. Therefore, a helmet-wearing detection algorithm based on the improved YOLOv8n model, PRE-YOLO, is proposed. First, we add small target detection layers and prune large target detection layers. The sophisticated algorithm considerably boosts the effectiveness of data manipulation while significantly reducing model parameters and size. Secondly, we introduce a convolutional module that integrates receptive field attention convolution and CA mechanisms into the backbone network, enhancing feature extraction capabilities by enhancing attention weights within both channel and spatial aspects. Lastly, we incorporate an EMA mechanism into the C2f module, which strengthens feature perception and captures more characteristic information while maintaining the same model parameter size. The experimental outcomes indicate that in comparison to the original model, the proposed PRE-YOLO model in this paper has improved by 1.3%, 1.7%, 2.2%, and 2.6% in terms of precision P, recall R, mAP@0.5, and mAP@0.5:0.95, respectively. At the same time, the number of model parameters has been reduced by 33.3%, and the model size has been reduced by 1.8 MB. Generalization experiments are conducted on the TWHD and EBHD datasets to further verify the versatility of the model. The research findings provide solutions for further improving the accuracy and efficiency of helmet-wearing detection on complex traffic roads, offering references for enhancing safety and intelligence in traffic.

Weakly-supervised temporal action localization using multi-branch attention weighting

Weakly-supervised temporal action localization using multi-branch attention weighting

Knowledge in attention assistant for improving generalization in deep teacher–student models

ABSTRACT Research on knowledge distillation has become active in deep neural networks. Knowledge distillation involves training a low-capacity model from a high-capacity model. However, when the capacities of the teacher and student models differ, it can result to poor learning and low generalization performance. We propose here a novel teacher assistant model called Knowledge in Attention Assistant. This model learns learning a discriminative representation of important regions and statistical information along with spatial and channel knowledge. Moreover, by using a triplet attention mechanism, the student model can learn both the inner and outer distribution of different categories, and also memorize the knowledge distribution of the teacher model. This alignment improves the effectiveness and generalization of knowledge distillation and reduces the capacity gap between the teacher and student models. The present model addresses feature inconsistency by adjusting the attention weight distribution based on the resemblance between the features of the teacher and student. The evaluation of the proposed teacher assistant method shows remarkable results. The student model outperforms the teacher model in terms of generalization performance, achieving improvements of 93.37% and 94.09% on CIFAR-10 and CIFAR-100 datasets, respectively. Furthermore, the proposed model enhances the F1-scores 91.98% on CIFAR-10 and 79.69% on CIFAR-100.

FCT-Net: A dual-encoding-path network fusing atrous spatial pyramid pooling and transformer for pavement crack detection

Cracks are a typical form of road damage, and accurate detection of cracks is of great significance for road maintenance work and ensuring traffic safety. Recently, computer vision has gradually been applied in the field of crack segmentation. However, there are still some extremely challenging problems in crack segmentation, such as complex backgrounds, information loss caused by pooling and convolution operations, and insufficient fusion of global and local semantic information. In response to the above problems, this paper proposes a dual-encoding-path network with U-Net architecture called FCT-Net, by fusing channel atrous spatial pyramid pooling (CASPP) and transformer. Specifically, CASPP obtains multi-scale receptive fields by incorporating spatial and channel attention, while refining and extracting local features. Meanwhile, we introduce long-short distance attention to construct a novel transformer with the prominent characteristic of interaction between local and global attention features. In addition, a residual convolution module is designed to enhance the local features of the transformer. Furthermore, we devise a multi-scale attention weight cross fusion module to aggregate the features of the dual encoding branch, for reducing information loss during downsampling and suppress background information. Eventually, we evaluate the performance of FCT-Net by experiments on three public datasets. Extensive experimental results show that FCT-Net achieves higher F1-score and mean intersection over union (mIoU) than state-of-the-art segmentation networks on the DeepCrack537 and CrackLS315 datasets. Meanwhile, it has excellent segmentation performance for cracks in complex scenes, with the highest recall, F1-score, and mIoU respectively as 85.64%, 81.67%, and 84.05% on the CrackTree260 dataset.

A method for image–text matching based on semantic filtering and adaptive adjustment

As image–text matching (a critical task in the field of computer vision) links cross-modal data, it has captured extensive attention. Most of the existing methods intended for matching images and texts explore the local similarity levels between images and sentences to align images with texts. Even though this fine-grained approach has remarkable gains, how to further mine the deep semantics between data pairs and focus on the essential semantics in data remains to be quested. In this work, a new semantic filtering and adaptive approach (FAAR) was proposed to ease the above problem. To be specific, the filtered attention (FA) module selectively focuses on typical alignments with the interference of meaningless comparisons eliminated. Next, the adaptive regulator (AR) further adjusts the attention weights of key segments for filtered regions and words. The superiority of our proposed method was validated by a number of qualitative experiments and analyses on the Flickr30K and MSCOCO data sets.

Precise Image-level Localization of Intracranial Hemorrhage on Head CT Scans with Deep Learning Models Trained on Study-level Labels.

"Just Accepted" papers have undergone full peer review and have been accepted for publication in Radiology: Artificial Intelligence. This article will undergo copyediting, layout, and proof review before it is published in its final version. Please note that during production of the final copyedited article, errors may be discovered which could affect the content. Purpose To develop a highly generalizable weakly supervised model to automatically detect and localize image- level intracranial hemorrhage (ICH) using study-level labels. Materials and Methods In this retrospective study, the proposed model was pretrained on the image-level RSNA dataset and fine-tuned on a local dataset using attention-based bidirectional long-short-term memory networks. This local training dataset included 10,699 noncontrast head CT scans from 7469 patients with ICH study-level labels extracted from radiology reports. Model performance was compared with that of two senior neuroradiologists on 100 random test scans using the McNemar test, and its generalizability was evaluated on an external independent dataset. Results The model achieved a positive predictive value (PPV) of 85.7% (95% CI: [84.0%, 87.4%]) and an AUC of 0.96 (95% CI: [0.96, 0.97]) on the held-out local test set (n = 7243, 3721 female) and 89.3% (95% CI: [87.8%, 90.7%]) and 0.96 (95% CI: [0.96, 0.97]), respectively, on the external test set (n = 491, 178 female). For 100 randomly selected samples, the model achieved performance on par with two neuroradiologists, but with a significantly faster (P < .05) diagnostic time of 5.04 seconds per scan (versus 86 seconds and 22.2 seconds for the two neuroradiologists, respectively). The model's attention weights and heatmaps visually aligned with neuroradiologists' interpretations. Conclusion The proposed model demonstrated high generalizability and high PPVs, offering a valuable tool for expedited ICH detection and prioritization while reducing false-positive interruptions in radiologists' workflows. ©RSNA, 2024.

Enhancing performance of transformer-based models in natural language understanding through word importance embedding

Transformer-based models have achieved state-of-the-art performance on natural language understanding (NLU) tasks by learning important token relationships through the attention mechanism. However, we observe that attention can become overly distributed during fine-tuning, failing to preserve the dependencies between meaningful tokens adequately. This phenomenon negatively affects the learning of token relationships in sentences. To overcome this issue, we propose a methodology that embeds the feature of word importance (WI) in the transformer-based models as a new layer, weighting the words according to their importance. Our simple yet powerful approach offers a general technique to boost transformer model capabilities on NLU tasks by mitigating the risk of attention dispersion during fine-tuning. Through extensive experiments on GLUE, SuperGLUE, and SQuAD benchmarks for pre-trained models (BERT, RoBERTa, ELECTRA, and DeBERTa), and MMLU, Big Bench Hard, and DROP benchmarks for the large language model, Llama2, we validate the effectiveness of our method in consistently enhancing performance across models with negligible overhead. Furthermore, we validate that our WI layer better preserves the dependencies between important tokens than standard fine-tuning by introducing a model classifying dependent tokens from the learned attention weights. The code is available at https://github.com/bigbases/WordImportance.

Design of an Improved Model for Lumpy Skin Disease Prediction Using Graph Neural Networks and Multiple Modal Data Fusion

An effective Lumpy Skin Disease prediction system mitigates economic losses and ensures animal health. Conventional approaches always fall short in terms of spatial-temporal analytics, very oft coming from a single source of data samples. To this regard, this research addresses the limitations by proposing a holistically integrated approach that puts together advanced machine learning techniques with enhanced disease prediction and management. In this paper, we provide a new framework that integrates graph neural networks for the fusion of graph-structured data with multivariables, attention mechanism, and self-supervised learning into compact feature representations. First, graph convolutional neural networks are utilized for spatial-temporal analysis. GCNs use node features, for instance, epidemiological data, and edge features in the form of connectivity data to generate node embeddings that capture complex dynamics of disease spread. It improved the prediction accuracy by up to 85%, greatly improving the spatial understanding in LSD dynamics. It basically fuses satellite image data, weather data, and livestock movement records through a multiple modal data fusion approach based on multiple head attention networks according to these embeddings. This will result in an enriched feature space in which the attention weights give insights into the relative importance of each data modality. This step improved the prediction accuracy by a further 5%, making the total prediction accuracy to be at 90%. Further fine-tuning of the model is done using self-supervised learning techniques through contrastive learning: SimCLR. This pre-paid feature representations using a large data set with unlabeled historic records on animal health data and environmental data samples. It improves model performance by up to 10% when fine-tuning the representations using LSD-specific data, thereby reducing reliance on labeled data and significantly improving the robustness and scalability of the model. This resulted in an overall accuracy of 95% for the fusion model, attained from the combination of these state-of-the-art techniques. In view of such comprehensive and diversified feature space, it showed very high resilience to adversarial attacks. Veterinary applications, in particular, support interpretability through attention mechanisms and node embeddings. Therefore, this work offers an accurate, robust, and interpretable solution to predict and manage Lumpy Skin Disease, contributing seriously to the domain of veterinary epidemiology.

Attention-based stackable graph convolutional network for multi-view learning

In multi-view learning, graph-based methods like Graph Convolutional Network (GCN) are extensively researched due to effective graph processing capabilities. However, most GCN-based methods often require complex preliminary operations such as sparsification, which may bring additional computation costs and training difficulties. Additionally, as the number of stacking layers increases in most GCN, over-smoothing problem arises, resulting in ineffective utilization of GCN capabilities. In this paper, we propose an attention-based stackable graph convolutional network that captures consistency across views and combines attention mechanism to exploit the powerful aggregation capability of GCN to effectively mitigate over-smoothing. Specifically, we introduce node self-attention to establish dynamic connections between nodes and generate view-specific representations. To maintain cross-view consistency, a data-driven approach is devised to assign attention weights to views, forming a common representation. Finally, based on residual connectivity, we apply an attention mechanism to the original projection features to generate layer-specific complementarity, which compensates for the information loss during graph convolution. Comprehensive experimental results demonstrate that the proposed method outperforms other state-of-the-art methods in multi-view semi-supervised tasks.

Modeling Sequences as Star Graphs to Address Over-Smoothing in Self-Attentive Sequential Recommendation

Self-attention (SA) mechanisms have been widely used in developing sequential recommendation (SR) methods, and demonstrated state-of-the-art performance. However, in this article, we show that self-attentive SR methods substantially suffer from the over-smoothing issue that item embeddings within a sequence become increasingly similar across attention blocks. As widely demonstrated in the literature, this issue could lead to a loss of information in individual items, and significantly degrade models’ scalability and performance. To address the over-smoothing issue, in this article, we view items within a sequence constituting a star graph and develop a method, denoted as \(\mathop{\mathtt{MSSG}}\limits\) , for SR. Different from existing self-attentive methods, \(\mathop{\mathtt{MSSG}}\limits\) introduces an additional internal node to specifically capture the global information within the sequence, and does not require information propagation among items. This design fundamentally addresses the over-smoothing issue and enables \(\mathop{\mathtt{MSSG}}\limits\) a linear time complexity with respect to the sequence length. We compare \(\mathop{\mathtt{MSSG}}\limits\) with eleven state-of-the-art baseline methods on six public benchmark datasets. Our experimental results demonstrate that \(\mathop{\mathtt{MSSG}}\limits\) significantly outperforms the baseline methods, with an improvement of as much as 10.10%. Our analysis shows the superior scalability of \(\mathop{\mathtt{MSSG}}\limits\) over the state-of-the-art self-attentive methods. Our complexity analysis and runtime performance comparison together show that \(\mathop{\mathtt{MSSG}}\limits\) is both theoretically and practically more efficient than self-attentive methods. Our analysis of the attention weights learned in SA-based methods indicates that on sparse recommendation data, modeling dependencies in all item pairs using the SA mechanism yields limited information gain, and thus, might not benefit the recommendation performance. Our source code and data are publicly accessible through GitHub .

Enhanced Detection and Recognition of Road Objects in Infrared Imaging Using Multi-Scale Self-Attention.

In infrared detection scenarios, detecting and recognizing low-contrast and small-sized targets has always been a challenge in the field of computer vision, particularly in complex road traffic environments. Traditional target detection methods usually perform poorly when processing infrared small targets, mainly due to their inability to effectively extract key features and the significant feature loss that occurs during feature transmission. To address these issues, this paper proposes a fast detection and recognition model based on a multi-scale self-attention mechanism, specifically for small road targets in infrared detection scenarios. We first introduce and improve the DyHead structure based on the YOLOv8 algorithm, which employs a multi-head self-attention mechanism to capture target features at various scales and enhance the model's perception of small targets. Additionally, to prevent information loss during the feature transmission process via the FPN structure in traditional YOLO algorithms, this paper introduces and enhances the Gather-and-Distribute Mechanism. By computing dependencies between features using self-attention, it reallocates attention weights in the feature maps to highlight important features and suppress irrelevant information. These improvements significantly enhance the model's capability to detect small targets. Moreover, to further increase detection speed, we pruned the network architecture to reduce computational complexity and parameter count, making the model suitable for real-time processing scenarios. Experiments on our self built infrared road traffic dataset (mainly including two types of targets: vehicles and people) show that compared with the baseline, our method achieves a 3.1% improvement in AP and a 2.5% increase in mAP on the VisDrone2019 dataset, showing significant enhancements in both detection accuracy and processing speed for small targets, with improved robustness and adaptability.

Residual attention based multi-label learning for apple leaf disease identification

Recent studies suggest that plant disease identification via machine learning approach is vital for preventing the spread of diseases. Identifying multiple diseases simultaneous on a single leaf is one of the most irritating issues in agricultural production. However, the existing approaches are difficult to meet the requirements of production practice in accuracy or interpretability. Here, we present residual attention based multi-label learning framework (RAMDI), a method for predicting apple leaf diseases in natural environment. Built upon an attention based multi-label learning framework, the channel and spatial attention mechanisms are investigated and embedded in residual network for multi-label disease prediction, which takes advantage of channel-wise and spatial-wise attention weights. Experimental results indicate that the RAMDI achieves 0.976 accuracy, 0.986 F-score, and 0.979 mAPs, outperforms the existing state-of-the-art apple leaf disease identification models. RAMDI not only predicts multi-disease on a single leaf simultaneously, but also reveals the interpretability among positive predictions that contribute most to identify the key features that are significant for the leaf diseases. This method achieves the following two achievements. Firstly, it provides a solution for detecting multiple diseases on a single leaf. Secondly, this approach gains an interpretable understanding for apple leaf disease identification.

A Novel Automated System for Early Diabetic Retinopathy Detection and Severity Classification

ABSTRACTDiabetes is a common and serious global disease that damages blood vessels in the eye, leading to vision loss. Early and accurate diagnosis of this issue is crucial to reduce the risk of visual impairment. The typical deep learning (DL) methods for diabetic retinopathy (DR) grading are often time‐consuming, resulting in unsatisfactory detection performance due to inadequate representation of lesion features. To overcome these challenges, this research proposes a new automated mechanism for detecting and classifying DR, aiming to identify DR severities and different stages. To figure out and capture feature characteristics from DR samples, a conjugated attention mechanism and vision transformer are utilized within a collective net model, which automatically generates feature maps for diagnosing DR. These extracted feature maps are then fused through the feature fusion function in a fused attention net model, calculating attention weights to produce the most powerful feature map. Finally, the DR cases are identified and discriminated using the kernel extreme learning machine (KELM) model. For evaluating DR severity, our work utilizes four different benchmark datasets: APTOS 2019, MESSIDOR‐2 dataset, DiaRetDB1 V2.1, and DIARETDB0 datasets. To illuminate data noise and unwanted variations, two preprocessing steps are carried out, which include contrast enhancement and illumination correction. The experimental results evaluated using well‐known indicators demonstrate that the suggested method achieves a higher accuracy of 99.63% compared to other baseline methods. This research contributes to the development of powerful DR screening techniques that are less time‐consuming and capable of automatically identifying DR severity levels at a premature level.

Distributed Regional Photovoltaic Power Prediction Based on Stack Integration Algorithm

With the continuous increase in the proportion of distributed photovoltaic power stations, the demand for photovoltaic power grid connection is becoming more and more urgent, and the requirements for the accuracy of regional distributed photovoltaic power forecasting are also increasing. A distributed regional photovoltaic power prediction model based on a stacked ensemble algorithm is proposed here. This model first uses a graph attention network (GAT) to learn the structural features and relationships between sub-area photovoltaic power stations, dynamically calculating the attention weights of the photovoltaic power stations to capture the global relationships and importance between stations, and selects representative stations for each sub-area. Subsequently, the CNN-LSTM-multi-head attention parallel multi-channel (CNN-LSTM-MHA (PC)) model is used as the basic model to predict representative stations for sub-areas by integrating the advantages of both the CNN and LSTM models. The predicted results are then used as new features for the input data of the meta-model, which finally predicts the photovoltaic power of the large area. Through comparative experiments at different seasons and time scales, this distributed regional approach reduced the MAE metric by a total of 22.85 kW in spring, 17 kW in summer, 30.26 kW in autumn, and 50.62 kW in winter compared with other models.

Human–object interaction detection via recycling of ground-truth annotations

Human–object interaction (HOI) detection is important to understand human-centric scenes, the hard core of which lies in learning the structure information from various types of relations. Existing works tackle it by introducing spatial context, extra knowledge or graph-based propagation based on the original hard labels. However, they still face challenges in dealing with action co-occurrence and complex HOIs. In this paper, we creatively propose to recycle the ground-truth annotations to get implied information for structure representations in HOIs. The action-aware closeness labeling (ACL) task is designed to capture the contextual information based on the statistic of action co-occurrence from the data source. Furthermore, we present a H-O relation graph supervision (RGS) to get more reliable relations in complicated scenes by constraining the attention weights of the H-O relation graph. Such a direct supervision on mutual relations is ignored in existing works. Experiments for HOI detection on the V-COCO and HICO-DET datasets indicate the superiority of the proposed method even without any extra knowledge.