the purpose of multimodal aspect sentiment analysis is to classify the sentimental polarities of the mentioned targets from graphic and textual data. However, previous approaches neglected fine-grained semantic associ...
详细信息
ISBN:
(纸本)9798350379860;9798350379877
the purpose of multimodal aspect sentiment analysis is to classify the sentimental polarities of the mentioned targets from graphic and textual data. However, previous approaches neglected fine-grained semantic associations between images and text, as well as the associations between image targets, resulting in limitation in multimodal aspect sentiment analysis. To tackle these difficulties, we suggest a fusion network model guided by image text similarity (ISGF). the fusion network utilizes aspect guidance attention and multimodal representation fusion module based on image text similarity to obtain effective multimodal fusion information. Align and fuse multimodal features by comparing learning methods, and assist in predicting the final answer. Our experimental outcomes on two public sentiment datasets demonstrate the feasibility and effectiveness of the model ISGF.
With rapid urbanization, fire accidents caused by electric vehicles in elevators have become increasingly frequent, posing a serious threat to the safety of residents. Traditional detection methods are inefficient and...
详细信息
ISBN:
(纸本)9798350379860;9798350379877
With rapid urbanization, fire accidents caused by electric vehicles in elevators have become increasingly frequent, posing a serious threat to the safety of residents. Traditional detection methods are inefficient and prone to both missing detection and misjudgment. this paper proposes an improved YOLOv8n-FS model based on YOLOv8, which combines the FasterNet network and SEAttention mechanism to improve the real-time detection performance of electric vehicles in elevators. FasterNet accelerates model inference through efficient feature extraction, while SEAttention enhances model feature representation and reduces missed and false detection. Experimental results demonstrate that the proposed YOLOv8n-FS model is significantly superior to the current mainstream methods in detection accuracy and speed, and has good practical application potential.
Detecting vehicles at low resolution is essential for traffic surveillance and autonomous *** high-resolution detection methods often struggle with low-resolution images, but super-resolution techniques can recover fi...
详细信息
ISBN:
(纸本)9798350379860;9798350379877
Detecting vehicles at low resolution is essential for traffic surveillance and autonomous *** high-resolution detection methods often struggle with low-resolution images, but super-resolution techniques can recover fine details, thereby significantly improving detection *** this paper, a target detection method that is based on YOLOv8s is presented, utilizing the ADown convolution module in place of the Conv module to reduce parameters and improve ***, the PSFM deep semantic fusion module is incorporated into the detection head to enhance feature fusion, while the Dattention mechanism is integrated into the C2f module to capture complex details. Experimental results demonstrate that the PSF3-YOLOv8 model outperforms other YOLO networks in key metrics, achieving improvements of 5.5% in Precision, 0.1% in Recall, 0.9% in mAP50, and 3.5% in mAP50-95.
In the Internet of Vehicles, vehicles lack reliable network access in scenarios such as unmanned areas. Satellite-Terrestrial Integrated Network (STIN) can provide reliable network access for vehicles in scenarios wit...
详细信息
ISBN:
(纸本)9798350379860;9798350379877
In the Internet of Vehicles, vehicles lack reliable network access in scenarios such as unmanned areas. Satellite-Terrestrial Integrated Network (STIN) can provide reliable network access for vehicles in scenarios without cellular network coverage. However, the existing satellite-terrestrial traffic scheduling methods often overlook the differences in traffic types, leading to limited quality of service (QoS) for STIN vehicles. To address this issue, we first model the attributes such as satellite-terrestrial visibility, filtering satellites that meet the minimum threshold requirements. Secondly, we design a customized QoS evaluation function and integrate it with switching costs and load balancing indices to create a reinforcement learning reward function, aiming to optimize satellite-terrestrial handover decisions by maximizing the reward. Finally, we develop an intelligent traffic allocation strategy based on reinforcement learning to ensure successful handover rates and onboard load balancing. Simulation results indicate that the proposed algorithm maintains high vehicle QoS and onboard load balance while reducing handover failure rates.
Deep generative models, such as Generative Adversarial Networks (GAN) and Variational Autoencoders (VAE), are widely used in collaborative filtering. they usually learn users' preferences for items directly from a...
详细信息
ISBN:
(纸本)9798350379860;9798350379877
Deep generative models, such as Generative Adversarial Networks (GAN) and Variational Autoencoders (VAE), are widely used in collaborative filtering. they usually learn users' preferences for items directly from a highly sparse user rating matrix (URM), and then recommend the top-N items to users. Due to the high sparsity of URM, GAN has limited ability to handle sparse data, while VAE's encoder can extract features. therefore, we propose a two-stage collaborative filtering framework based on variational autoencoders and generative adversarial networks, named VCFGAN. It first uses two VAEs to extract features from URM and side information (SI), and then uses the extracted latent vectors to train the GAN network. To evaluate the performance of our proposed VCFGAN model, some experiments are conducted on two real datasets, and the experimental results show that our model outperforms other representative models.
this article proposes a novel repetition counting prediction model that improves the accuracy of estimating the frequency of repetitive actions in different scenarios and generalization capability across different dat...
详细信息
ISBN:
(纸本)9798350379860;9798350379877
this article proposes a novel repetition counting prediction model that improves the accuracy of estimating the frequency of repetitive actions in different scenarios and generalization capability across different datasets by incorporating Pose Estimation and Focused Linear Attention. Our model uses focused linear attention to calculate the spatiotemporal similarity between pose estimations of target objects in video frames, highlighting the motion behavior of foreground objects,reducing the influence of background image variations on the results and enhancing the model's generalization ability. Experimental results show that model effectively reduces computational costs for accurately predicting the number of repetitions. the focused linear attention mechanism contributes to a significant reduction in model FLOPs compared to others and a slight increase in OBO metric. Otherwise the model not only demonstrates excellent performance on human activity datasets but also performs well on animal datasets, showcasing its generalization capabilities.
In order to measure the workload of pilots under the conditions of day and night alternation, the task load analysis method of the improved VACP model was used to reconstruct the cognitive natural decay function of th...
详细信息
ISBN:
(纸本)9798350379860;9798350379877
In order to measure the workload of pilots under the conditions of day and night alternation, the task load analysis method of the improved VACP model was used to reconstruct the cognitive natural decay function of the steady-state process in the SAFTE model, and a pilot workload assessment model that comprehensively considers resource decay, sleep time, circadian rhythm and task influencing factors was established. In order to verify the accuracy of the model, a high-workload simulation flight mission was designed, and a validation data set was obtained. the calculation results of the workload assessment model established in this paper were compared withthe experimental data. the results verified the accuracy of the model in identifying and locating high-workload task periods.
Cognitive task analysis methods have been extensively applied across various fields since the 1980s. Among these, the GOMS (Goals, Operators, Methods, Selection rules) model stands out as one of the most widely adopte...
详细信息
ISBN:
(纸本)9798350379860;9798350379877
Cognitive task analysis methods have been extensively applied across various fields since the 1980s. Among these, the GOMS (Goals, Operators, Methods, Selection rules) model stands out as one of the most widely adopted cognitive task analysis models in the domain of human-machine ergonomics. However, the GOMS model traditionally evaluates only two parameters during the analysis process: learning time and execution time. To enhance this model, this paper introduces the GOMSL-HCR (Goals, Operators, Methods, Selection Rules Language - Human Cognitive Reliability) model, which incorporates an improved HCR (Human Cognitive Reliability) model. the GOMSL-HCR model is designed to predict both task execution time and the probability of failure. the accuracy of the model's predictions is validated through experimental testing.
the dearth of labeled Tangut ancient book pages severely hampers the development of accurate text detection models. To mitigate this issue, we introduce a lightweight few-shot object detection model Tangut-YOLOv8. Dra...
详细信息
ISBN:
(纸本)9798350379860;9798350379877
the dearth of labeled Tangut ancient book pages severely hampers the development of accurate text detection models. To mitigate this issue, we introduce a lightweight few-shot object detection model Tangut-YOLOv8. Drawing inspiration from the Two-stage Fine-tuning Approach(TFA), we also employs a similar strategy. the first stage utilizes a large corpus of Chinese ancient book pages to imbue the model with general text detection capabilities. In the subsequent fine-tuning stage, we fine-tune the model to adapt to the specific characteristics of the Tangut script with a metric component. Experimental results reveal that our model outperforms several few-shot object detection models on the Tangut ancient book page dataset. this approach offers an innovative solution for text detection in ancient book pages, facilitating the preservation and scholarly analysis of this invaluable cultural heritage.
the complexity of smoke shapes, textures, and colors poses a challenge for accurate smoke recognition. To tackle this challenge, we propose a two-channel smoke recognition algorithm incorporating the Convolutional Blo...
详细信息
ISBN:
(纸本)9798350379860;9798350379877
the complexity of smoke shapes, textures, and colors poses a challenge for accurate smoke recognition. To tackle this challenge, we propose a two-channel smoke recognition algorithm incorporating the Convolutional Block Attention Module (CBAM). the algorithm first employs a dual-channel architecture combining Skip Connection-based Neural Network (SCNN) and Selective-based Batch Normalization Network (SBNN) to effectively capture both basic and detailed features. Subsequently, CBAM is embedded to suppress background interference and enhance smoke features at the output of the two-channel network. Experiments using publicly available smoke datasets demonstrate that the proposed algorithm achieves a smoke recognition accuracy of 99.2%, outperforming existing other methods.
暂无评论