Modulation Of Aggregation Research Articles

MoMFormer: Mixture of modality transformer model for vegetation extraction under shadow conditions

Accurate estimation of fractional vegetation coverage (FVC) is essential for assessing the ecological environment and acquiring ecological information. However, under natural lighting conditions, shadows in vegetation scenes can easily lead to confusion between shadowed vegetation and shadowed soil, leading to misclassification and omission errors. This issue limits the precision of both vegetation extraction and FVC estimation. To address this challenge, this study introduces a novel deep learning model, the Mixture of Modality Transformer (MoMFormer), which is specifically designed to mitigate shadow interference in vegetation extraction. Our model uses the Swin-transformer V2 as a feature extractor, effectively capturing vegetation features from a dual-modality (regular-exposure RGB and high dynamic range HDR) dataset. A dynamic aggregation module (DAM) is integrated to adaptively blend the most relevant vegetation features. We selected several state-of-the-art (SOTA) methods and conducted extensive experiments using a self-annotated dataset featuring diverse vegetation–soil scenes and compare our model with several state-of-the-art methods. The results demonstrate that MoMFormer achieves an accuracy of 89.43 % on the HDR-RGB dual-modality dataset, with an FVC accuracy of 87.57 %, outperforming other algorithms and demonstrating high vegetation extraction accuracy and adaptability under natural lighting conditions. This research offers new insights into accurate vegetation information extraction in naturally lit environments with shadows, providing robust technical support for high-precision validation of vegetation coverage products and algorithms based on multimodal data. The code and datasets used in this study are publicly available at https://github.com/hhhxiaohe/MoMFormer.

RTIA-Mono: Real-time lightweight self-supervised monocular depth estimation with global-local information aggregation

Self-supervised monocular depth estimation has attracted significant attention in computer vision, especially for applications like autonomous driving and robotics. Recently, CNNs and Transformers have achieved tremendous success in this task. However, existing research primarily focuses on improving estimation accuracy, increasing model complexity poses challenges for deployment on edge computing devices. Shallow CNNs aid lightweight network construction but suffer limited receptive fields, hindering fusion of local geometric features and global semantic information. To address these issues, we propose an efficient real-time lightweight self-supervised architecture, RTIA-Mono, for monocular depth estimation. Firstly, we design a cross-stage feature fusion structure promoting feature aggregation and fusion across stages. Secondly, in each stage, we propose a Global Local Information Aggregation (GLIA) module integrating advantages of CNNs and Transformers to aggregate local and global features. Additionally, we introduce a Directional Feature Enhancement (DFE) module supplementing spatial structure information to mitigate spatial information loss from downsampling. Through sophisticated design, the proposed approach outperforms state-of-the-art methods on KITTI benchmark with the least parameters, and achieves a good balance between accuracy, complexity and inference speed. Furthermore, RTIA-Mono demonstrates excellent generalization on other datasets.

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.

MLFA-Net: multi-level feature-aggregated network for semantic change detection in remote sensing images

ABSTRACT The rapid advancement in Earth observation technologies has improved the acquisition of precise data on terrestrial changes. However, traditional binary change detection (BCD) fails to satisfy the complex demands of contemporary applications. In contrast, semantic change detection (SCD) can identify changed areas and further detect the types of changes in ground objects. Nonetheless, current SCD approaches pose significant challenges, including the inadequate extraction and utilization of multi-level features and limited synergistic integration of related tasks. This paper proposes a multi-level feature-aggregated network (MLFA-Net), which decomposes the SCD task into two related subtasks: ground object classification and BCD of bi-temporal images. To optimize the classification performance, MLFA-Net incorporates feature alignment and cross-level feature aggregation modules that operate between the encoders and decoders, achieving effective extraction, aggregation, and utilization of multi-level image features. MLFA-Net also integrates symmetric transform feature aggregation modules between classification and BCD branches, leveraging prior knowledge to facilitate cross-task transmission of multi-level information and further optimize SCD results. MLFA-Net achieved Scores of 0.6214 and 0.3581 on Landsat-SCD and SECOND datasets, respectively, with improvements of at least 0.0402 and 0.0035 compared to other methods. Quantitative and qualitative experiments validate the efficacy and superiority of the proposed MLFA-Net in SCD tasks.

A Multi-Scale Deep Learning Algorithm for Enhanced Forest Fire Danger Prediction Using Remote Sensing Images

Forest fire danger prediction models often face challenges due to spatial and temporal limitations, as well as a lack of universality caused by regional inconsistencies in fire danger features. To address these issues, we propose a novel algorithm, squeeze-excitation spatial multi-scale transformer learning (SESMTML), which is designed to extract multi-scale fire danger features from remote sensing images. SESMTML includes several key modules: the multi-scale deep feature extraction module (MSDFEM) captures global visual and multi-scale convolutional features, the multi-scale fire danger perception module (MFDPM) explores contextual relationships, the multi-scale information aggregation module (MIAM) aggregates correlations of multi-level fire danger features, and the fire danger level fusion module (FDLFM) integrates the contributions of global and multi-level features for predicting forest fire danger. Experimental results demonstrate the model’s significant superiority, achieving an accuracy of 83.18%, representing a 22.58% improvement over previous models and outperforming many widely used deep learning methods. Additionally, a detailed forest fire danger prediction map was generated using a test study area at the junction of the Miyun and Pinggu districts in Beijing, further confirming the model’s effectiveness. SESMTML shows strong potential for practical application in forest fire danger prediction and offers new insights for future research utilizing remote sensing images.

Snow Detection in Gaofen-1 Multi-Spectral Images Based on Swin-Transformer and U-Shaped Dual-Branch Encoder Structure Network with Geographic Information

Snow detection is imperative in remote sensing for various applications, including climate change monitoring, water resources management, and disaster warning. Recognizing the limitations of current deep learning algorithms in cloud and snow boundary segmentation, as well as issues like detail snow information loss and mountainous snow omission, this paper presents a novel snow detection network based on Swin-Transformer and U-shaped dual-branch encoder structure with geographic information (SD-GeoSTUNet), aiming to address the above issues. Initially, the SD-GeoSTUNet incorporates the CNN branch and Swin-Transformer branch to extract features in parallel and the Feature Aggregation Module (FAM) is designed to facilitate the detail feature aggregation via two branches. Simultaneously, an Edge-enhanced Convolution (EeConv) is introduced to promote snow boundary contour extraction in the CNN branch. In particular, auxiliary geographic information, including altitude, longitude, latitude, slope, and aspect, is encoded in the Swin-Transformer branch to enhance snow detection in mountainous regions. Experiments conducted on Levir_CS, a large-scale cloud and snow dataset originating from Gaofen-1, demonstrate that SD-GeoSTUNet achieves optimal performance with the values of 78.08%, 85.07%, and 92.89% for IoU_s, F1_s, and MPA, respectively, leading to superior cloud and snow boundary segmentation and thin cloud and snow detection. Further, ablation experiments reveal that integrating slope and aspect information effectively alleviates the omission of snow detection in mountainous areas and significantly exhibits the best vision under complex terrain. The proposed model can be used for remote sensing data with geographic information to achieve more accurate snow extraction, which is conducive to promoting the research of hydrology and agriculture with different geospatial characteristics.

NLA-GNN: Non-local information aggregated graph neural network for heterogeneous graph embedding

Heterogeneous graphs are ubiquitous in the real world. Recent methods aim to obtain meaningful low-dimensional node embeddings from heterogeneous graphs to facilitate downstream applications. However, most existing methods tend to consider the local information but ignore the non-local information. This paper proposes a novel Non-Local Information Aggregated Graph Neural Network (NLA-GNN) that aggregate not only the local information from neighbor nodes but also non-local information from distant nodes. Specifically, Local aggregation modules in NLA-GNN utilize the attention mechanism to generate potentially valuable metapaths and use them to aggregate local information. In contrast, non-local aggregation modules adopt a two-step approach, and each step uses an attention-guided method to sort nodes into node sequences and aggregate information with the methods designed for sequential data. Experiment results on three heterogeneous graph datasets demonstrate the performance of NLA-GNN over state-of-the-arts and the necessity of non-local aggregation in heterogeneous graphs.

Transition Metal Complexes as Therapeutics: A New Frontier in Combatting Neurodegenerative Disorders through Protein Aggregation Modulation.

Neurodegenerative disorders (NDDs) are a class of debilitating diseases that progressively impair the protein structure and result in neurological dysfunction in the nervous system. Among these disorders, Alzheimer's disease (AD), prion diseases such as Creutzfeldt-Jakob disease (CJD), and Parkinson's disease (PD) are caused by protein misfolding and aggregation at the cellular level. In recent years, transition metal complexes have gained significant attention for their potential applications in diagnosing, imaging, and curing these NDDs. These complexes have intriguing possibilities as therapeutics due to their diverse ligand systems and chemical properties and can interact with biological systems with minimal detrimental effects. This review focuses on the recent progress in transition metal therapeutics as a new era of hope in the battle against AD, CJD, and PD by modulating protein aggregation in vitro and in vivo. It may shed revolutionary insights into unlocking new opportunities for researchers to develop metal-based drugs to combat NDDs.

CafeNet: A Novel Multi‐Scale Context Aggregation and Multi‐Level Foreground Enhancement Network for Polyp Segmentation

ABSTRACTThe detection of polyps plays a significant role in colonoscopy examinations, cancer diagnosis, and early patient treatment. However, due to the diversity in the size, color, and shape of polyps, as well as the presence of low image contrast with the surrounding mucosa and fuzzy boundaries, precise polyp segmentation remains a challenging task. Furthermore, this task requires excellent real‐time performance to promptly and efficiently present predictive results to doctors during colonoscopy examinations. To address these challenges, a novel neural network, called CafeNet, is proposed in this paper for rapid and accurate polyp segmentation. CafeNet utilizes newly designed multi‐scale context aggregation (MCA) modules to adapt to the extensive variations in polyp morphology, covering small to large polyps by fusing simplified global contextual information and local information at different scales. Additionally, the proposed network utilizes newly designed multi‐level foreground enhancement (MFE) modules to compute and extract differential features between adjacent layers and uses the prediction output from the adjacent lower‐layer decoder as a guidance map to enhance the polyp information extracted by the upper‐layer encoder, thereby improving the contrast between polyps and the background. The polyp segmentation performance of the proposed CafeNet network is evaluated on five benchmark public datasets using six evaluation metrics. Experimental results indicate that CafeNet outperforms the state‐of‐the‐art networks, while also exhibiting the least parameter count along with excellent real‐time operational speed.

Graph Representation Learning Based on Specific Subgraphs for Biomedical Interaction Prediction.

Discovering the novel associations of biomedical entities is of great significance and can facilitate not only the identification of network biomarkers of disease but also the search for putative drug targets.Graph representation learning (GRL) has incredible potential to efficiently predict the interactions from biomedical networks by modeling the robust representation for each node.> However, the current GRL-based methods learn the representation of nodes by aggregating the features of their neighbors with equal weights. Furthermore, they also fail to identify which features of higher-order neighbors are integrated into the representation of the central node. In this work, we propose a novel graph representation learning framework: a multi-order graph neural network based on reconstructed specific subgraphs (MGRS) for biomedical interaction prediction. In the MGRS, we apply the multi-order graph aggregation module (MOGA) to learn the wide-view representation by integrating the multi-hop neighbor features. Besides, we propose a subgraph selection module (SGSM) to reconstruct the specific subgraph with adaptive edge weights for each node. SGSM can clearly explore the dependency of the node representation on the neighbor features and learn the subgraph-based representation based on the reconstructed weighted subgraphs. Extensive experimental results on four public biomedical networks demonstrate that the MGRS performs better and is more robust than the latest baselines.

Multi-scale convolution enhanced transformer for multivariate long-term time series forecasting

In data analysis and forecasting, particularly for multivariate long-term time series, challenges persist. The Transformer model in deep learning methods has shown significant potential in time series forecasting. The Transformer model’s dot-product attention mechanism, however, due to its quadratic computational complexity, impairs training and forecasting efficiency. In addition, the Transformer architecture has limitations in modeling local features and dealing with multivariate cross-dimensional dependency relationship. In this article, a Multi-Scale Convolution Enhanced Transformer model (MSCformer) is proposed for multivariate long-term time series forecasting. As an alternative to modeling the time series in its entirety, a segmentation strategy is designed to convert the input original series into segmented forms with different lengths, then process time series segments using a new constructed multi-Dependency Aggregation module. This multi-Scale segmentation approach reduces the computational complexity of the attention mechanism part in subsequent models, and for each segment of length corresponds to a specific time scale, it also ensures that each segment retains the semantic information of the data sequence level, thereby comprehensively utilizing the multi-scale information of the data while more accurately capturing the real dependency of the time series. The Multi-Dependence Aggregate module captures both cross-temporal and cross-dimensional dependencies of multivariate long-term time series and compensates for local dependencies within the segments thereby captures local series features comprehensively and addressing the issue of insufficient information utilization. MSCformer synthesizes dependency information extracted from various temporal segments at different scales and reconstructs future series using linear layers. MSCformer exhibits higher forecasting accuracy, outperforming existing methods in multiple domains including energy, transportation, weather, electricity, disease and finance.

Cross-view discrepancy-dependency network for volumetric medical image segmentation

The limited data poses a crucial challenge for deep learning-based volumetric medical image segmentation, and many methods have tried to represent the volume by its subvolumes (i.e., multi-view slices) for alleviating this issue. However, such methods generally sacrifice inter-slice spatial continuity. Currently, a promising avenue involves incorporating multi-view information into the network to enhance volume representation learning, but most existing studies tend to overlook the discrepancy and dependency across different views, ultimately limiting the potential of multi-view representations. To this end, we propose a cross-view discrepancy-dependency network (CvDd-Net) to task with volumetric medical image segmentation, which exploits multi-view slice prior to assist volume representation learning and explore view discrepancy and view dependency for performance improvement. Specifically, we develop a discrepancy-aware morphology reinforcement (DaMR) module to effectively learn view-specific representation by mining morphological information (i.e., boundary and position of object). Besides, we design a dependency-aware information aggregation (DaIA) module to adequately harness the multi-view slice prior, enhancing individual view representations of the volume and integrating them based on cross-view dependency. Extensive experiments on four medical image datasets (i.e., Thyroid, Cervix, Pancreas, and Glioma) demonstrate the efficacy of the proposed method on both fully-supervised and semi-supervised tasks.

A semantic guidance-based fusion network for multi-label image classification

Multi-label image classification (MLIC), a fundamental task assigning multiple labels to each image, has been seen notable progress in recent years. Considering simultaneous appearances of objects in the physical world, modeling object correlations is crucial for enhancing classification accuracy. This involves accounting for spatial image feature correlation and label semantic correlation. However, existing methods struggle to establish these correlations due to complex spatial location and label semantic relationships. On the other hand, regarding the fusion of image feature relevance and label semantic relevance, existing methods typically learn a semantic representation in the final CNN layer to combine spatial and label semantic correlations. However, different CNN layers capture features at diverse scales and possess distinct discriminative abilities. To address these issues, in this paper we introduce the Semantic Guidance-Based Fusion Network (SGFN) for MLIC. To model spatial image feature correlation, we leverage the advanced TResNet architecture as the backbone network and employ the Feature Aggregation Module for capturing global spatial correlation. For label semantic correlation, we establish both local and global semantic correlation. We further enrich model features by learning semantic representations across multiple convolutional layers. Our method outperforms current state-of-the-art techniques on PASCAL VOC (2007, 2012) and MS-COCO datasets.

Multi-scale hybrid attention aggregation networks for multi-modal monitoring in laser-induced thermal-crack processing

In recent years, the rapid development of deep learning has led to an extensive amount of research into machining condition monitoring technologies. However, the complexity of multi-modal signals present in actual machining scenarios poses a significant challenge to signal fusion, and most models find it difficult to balance the relationship between performance and deployment cost. In this research, a multi-scale hybrid attention aggregation network (MHAAN) is proposed for multi-modal monitoring in laser-induced thermal-crack processing. Based on the feature distribution of visual signals and acoustic emission (AE) signals, the multi spatial scale aggregation module (MSSAM) and Multi frequency domain aggregation module (MFDAM) are designed. A cross-modal dot-product interaction module is employed to facilitate multilevel interactive fusion of the extracted feature vectors. This module enables feature transfer across different modalities, and then extracts the multi-model fusion feature. The practicality of MHAAN has been validated using a dataset obtained from the process of glass cutting with laser-induced thermal-crack propagation (LITP). Compared with commonly used models, the superiority of MHAAN in multi-modal feature extraction and fusion is demonstrated. Finally, the effectiveness of the module design based on prior knowledge is proven through ablation experiments, improving the interpretability of the model.

Multi-view human pose and shape estimation via mesh-aligned voxel interpolation

Although multi-view human pose and shape regression methods have information from other views for complementing and correcting, existing ones still have its own drawback of not fully taking advantage of multi-view setup. Thus they are far from efficiently aligning and merging features in different views. In order to tackle these problems, we propose a multi-view framework where features from all views are well aligned and merged through multi-view voxel aggregation with inverse projection. Our framework highlights three major characteristics. Firstly, we use a multi-view volumetric aggregation module for better prediction by exploiting various information in different-scale feature maps. Secondly, in our framework, instead of using all voxels, a mesh-aligned voxel selection module is proposed to make effective prediction by eliminating redundant background voxels. Lastly, the framework further improves the performance of human body parametric modeling by adopting a dual-branch strategy, where one branch for parametric human model prediction and the other for 3D keypoints prediction. Their mutual influence is critical to the improvement for both tasks. Additionally, we find the scarcity of datasets also hinders the development of multi-view methods, so we propose a approach for creating occlusion datasets specifically for multi-view occlusion case. Experimental results verify the effectiveness of the proposed framework on two benchmarks, Human3.6M and MPI-INF-3DHP.

EHOPN: A novel enhanced high-order pooling-based network for industrial fault detection

Recently, deep learning algorithms have been successfully applied to industrial fault detection because they are better at automatically extracting complex features and processing high-dimensional data than traditional methods. However, most existing deep learning-based fault detection methods only concentrate on extracting features from industrial process data without considering the crucial long-term temporal features and higher-order statistical information. To address this challenge, we proposed a novel enhanced higher-order pooling-based network (EHOPN) for industrial fault detection. First, the data pre-processing of the network is presented to capture the dynamic features of time-series process data and unify the high-dimensional data scale. Second, the EHOPN utilizes channel and temporal second-order pooling techniques to gather temporal and channel statistics information, facilitating the backbone network’s ability to capture complex inter-dependencies and long-term dynamics. Additionally, the high-order feature aggregation module is presented to aggregate global and local features, enhancing the network’s generalization ability. The proposed industrial fault detection approach is evaluated on the Tennessee Eastman benchmark and a real-world heavy-plate production process. Experimental results show that the proposed method is significantly better than comparison models in four evaluation metrics: accuracy, precision, recall, and F1-score, further proving the effectiveness of EHOPN.

A visible-infrared person re-identification method based on meta-graph isomerization aggregation module

Due to different imaging principles of visible-infrared cameras, there are modal differences between similar person images. For visible-infrared person re-identification (VI-ReID), existing works focus on extracting and aligning cross-modal global descriptors in the shared feature space, while ignoring local variations and graph-structural correlations in cross-modal image pairs. In order to bridge the modal differences with graph-structure between key-parts, a Meta-Graph Isomerization Aggregation Module (MIAM) is proposed, which includes Meta-Graph Node Isomerization Module (MNI) and Dual Aggregation Module (DA). In order to completely describe discriminative in-graph local features, MNI establishes meta-secondary cyclic isomorphism relations of in-graph local features by using a multi-branch embedding generation mechanism. Then, the local features not only contain the limited information of the fixed regions, but also benefit from the neighboring regions. Meanwhile, the secondary node generation process considers similar and different nodes of pedestrian graph structure to reduce the interference of identity differences. In addition, the Dual Aggregation (DA) module combines spatial self-attention and channel self-attention to establish the interdependence of modal heterogeneous graph structure pairs and achieve inter-modal feature aggregation. To match heterogeneous graph structures, a Node Center Joint Mining Loss (NCJM Loss) is proposed to constrain the distance between the node centers of heterogeneous graphs. The experiments performed on the SYSU-MM01, RegDB and LLCM public datasets demonstrate that the proposed method performs excellently in VI-ReID.

DAEiS-Net: Deep Aggregation Network with Edge Information Supplement for Tunnel Water Stain Segmentation.

Tunnel disease detection and maintenance are critical tasks in urban engineering, and are essential for the safety and stability of urban transportation systems. Water stain detection presents unique challenges due to its variable morphology and scale, which leads to insufficient multiscale contextual information extraction and boundary information loss in complex environments. To address these challenges, this paper proposes a method called Deep Aggregation Network with Edge Information Supplement (DAEiS-Net) for detecting tunnel water stains. The proposed method employs a classic encoder-decoder architecture. Specifically, in the encoder part, a Deep Aggregation Module (DAM) is introduced to enhance feature representation capabilities. Additionally, a Multiscale Cross-Attention Module (MCAM) is proposed to suppress noise in the shallow features and enhance the texture information of the high-level features. Moreover, an Edge Information Supplement Module (EISM) is designed to mitigate semantic gaps across different stages of feature extraction, improving the extraction of water stain edge information. Furthermore, a Sub-Pixel Module (SPM) is proposed to fuse features at various scales, enhancing edge feature representation. Finally, we introduce the Tunnel Water Stain Dataset (TWS), specifically designed for tunnel water stain segmentation. Experimental results on the TWS dataset demonstrate that DAEiS-Net achieves state-of-the-art performance in tunnel water stain segmentation.

Dual cross perception network with texture and boundary guidance for camouflaged object detection

Camouflaged object detection (COD) is a task needs to segment objects that subtly blend into their surroundings effectively. Edge and texture information of the objects can be utilized to reveal the edges of camouflaged objects and detect texture differences between camouflaged objects and the surrounding environment. However, existing methods often fail to fully exploit the advantages of these two types of information. Considering this, our paper proposes an innovative Dual Cross Perception Network (DCPNet) with texture and boundary guidance for camouflaged object detection. DCPNet consists of two essential modules, namely Dual Cross Fusion Module (DCFM) and the Subgroup Aggregation Module (SAM). DCFM utilizes attention techniques to emphasize the information that exists in edges and textures by cross-fusing features of the edge, texture, and basic RGB image, which strengthens the ability to capture edge information and texture details in image analysis. SAM gives varied weights to low-level and high-level features in order to enhance the comprehension of objects and scenes of various sizes. Several experiments have demonstrated that DCPNet outperforms 13 state-of-the-art methods on four widely used assessment metrics.

TS-BEV: BEV object detection algorithm based on temporal-spatial feature fusion

In order to accurately identify occluding targets and infer the motion state of objects, we propose a Bird’s-Eye View Object Detection Network based on Temporal-Spatial feature fusion (TS-BEV), which replaces the previous multi-frame sampling method by using the cyclic propagation mode of historical frame instance information. We design a new Temporal-Spatial feature fusion attention module, which fully integrates temporal information and spatial features, and improves the inference and training speed. In response to realize multi-frame feature fusion across multiple scales and views, we propose an efficient Temporal-Spatial deformable aggregation module, which performs feature sampling and weighted summation from multiple feature maps of historical frames and current frames, and makes full use of the parallel computing capabilities of GPUs and AI chips to further improve efficiency. Furthermore, in order to solve the lack of global inference in the context of temporal-spatial fusion BEV features and the inability of instance features distributed in different locations to fully interact, we further design the BEV self-attention mechanism module to perform global operation of features, enhance global inference ability and fully interact with instance features. We have carried out extensive experimental experiments on the challenging BEV object detection nuScenes dataset, quantitative results show that our method achieves excellent performance of 61.5% mAP and 68.5% NDS in camera-only 3D object detection tasks, and qualitative results show that TS-BEV can effectively solve the problem of 3D object detection in complex traffic background with lack of light at night, with good robustness and scalability.