Dilated Convolutional Layers Research Articles

ARFNet: Attention-Oriented Refinement and Fusion Network for Light Field Salient Object Detection

Light field salient object detection plays an important role in computer vision tasks because of the excellent performance on challenging scenes. However, the imprecise extraction of saliency features from light field imaging puts a negative impact on the accuracy of saliency results, and the invalid fusion of these extracted features makes saliency results lose details, further cutting down their accuracy. In this article, to effectively extract and fuse saliency features of light field imaging, we propose an attention-oriented refinement and fusion network to solve these problems, which mainly includes the attention-oriented refinement module (ARM) and the attention-oriented fusion module (AFM). Specifically, ARM can precisely refine similar features between focal slices and all-focus image, and fully extract abundant structural characteristics of the light field. Subsequently, AFM can efficiently fuse those features extracted by ARM to obtain saliency results with detail information. Furthermore, different dilated convolution layers embedded with ARM and AFM are leveraged to further capture complete saliency results with detail information. Experimental results demonstrate that our proposed method outperforms other 20 state-of-the-art methods and achieves Top-1 accuracies on three light field datasets.

Global–local multi-stage temporal convolutional network for cataract surgery phase recognition

BackgroundSurgical video phase recognition is an essential technique in computer-assisted surgical systems for monitoring surgical procedures, which can assist surgeons in standardizing procedures and enhancing postsurgical assessment and indexing. However, the high similarity between the phases and temporal variations of cataract videos still poses the greatest challenge for video phase recognition.MethodsIn this paper, we introduce a global–local multi-stage temporal convolutional network (GL-MSTCN) to explore the subtle differences between high similarity surgical phases and mitigate the temporal variations of surgical videos. The presented work consists of a triple-stream network (i.e., pupil stream, instrument stream, and video frame stream) and a multi-stage temporal convolutional network. The triple-stream network first detects the pupil and surgical instruments regions in the frame separately and then obtains the fine-grained semantic features of the video frames. The proposed multi-stage temporal convolutional network improves the surgical phase recognition performance by capturing longer time series features through dilated convolutional layers with varying receptive fields.ResultsOur method is thoroughly validated on the CSVideo dataset with 32 cataract surgery videos and the public Cataract101 dataset with 101 cataract surgery videos, outperforming state-of-the-art approaches with 95.8% and 96.5% accuracy, respectively.ConclusionsThe experimental results show that the use of global and local feature information can effectively enhance the model to explore fine-grained features and mitigate temporal and spatial variations, thus improving the surgical phase recognition performance of the proposed GL-MSTCN.

AFSPNet: an adaptive feature selection pyramid network for efficient object detection in remote sensing images

In recent studies, remote sensing object detection methods based on deep learning have emerged as a primary concern in environmental monitoring, military investigation, and hazard response. However, many difficulties, such as complex backgrounds, dense target quantities, large-scale variations, and non-uniform distribution, lead to many parameters and complex network structures, thus limiting the accuracy of the detector and slowing the inference speed. To address these issues, we propose a lightweight and efficient object detector for remote sensing images. First, an asymmetric convolution with the visual attention mechanism is reconstructed to decrease the complexity and strengthen the feature representation ability. Then, an adaptive feature selection structure is designed to extract discriminative feature information, which can adaptively model the shapes of objects by introducing deformable convolution to obtain a stronger geometric feature representation. To reduce information loss across different channels and spatial locations, a hybrid receptive field module is also proposed to increase the receptive field model by mixing dilated convolutional layers with different dilation rates. Finally, experimental results on the DIOR dataset show that our approach significantly improves detection accuracy and running speed.

Simple dilated convolutional neural network for quantitative modeling based on near infrared spectroscopy techniques

A simple deep convolutional neural network architecture had been developed and applied to the establishment of a quantitative model based on near infrared spectroscopy techniques. The network architecture only contained general convolutional layers and dilated convolutional layers. The determination coefficient was used as a criterion to stop the training process of the neural network. Three data sets were used to test the performance of the neural network. The major results are (1) quantitative model can be established within one million iterations; (2) the quantitative model established on one spectrometer can be applied to other spectrometers of same manufacturer and even the same kind of spectrometer of different manufacturer; (3) simultaneous quantitative analysis of four components in grain samples. This study provided a new strategy to apply techniques of deep neural network in quantitative analysis.

CASDD: Automatic Surface Defect Detection Using a Complementary Adversarial Network

Surface defect detection (SDD) plays an extremely important role in the manufacturing stage of products. However, this is a fundamental yet challenging task, mainly because the intraclass defects have large differences in shape and distribution, and low contrast between the object regions and background, and it is difficult to adapt to other materials. To address this problem, we propose a complementary adversarial network-driven SDD (CASDD) framework to automatically and accurately identify various types of texture defects. Specifically, CASDD consists of an encoding–decoding segmentation module with a specially designed loss measurement and a novel complementary discriminator mechanism. In addition, to model the defect boundaries and enhance the feature representation, the dilated convolutional (DC) layers with different rates and edge detection (ED) blocks are also incorporated into CASDD. Moreover, a complementary discrimination strategy is proposed, which employs two independent yet complementary discriminator modules to optimize the segmentation module more effectively. One discriminator identifies contextual features of the object regions in the input defect images, while the other discriminator focuses on edge detail differences between the ground truth and the segmented image. To obtain more edge information during training, a new composite loss measurement containing edge information and structural features is designed. Experimental results show that CASDD can be suitable for defect detection on four real-world and one artificial defect database, and its detection accuracy is significantly better than the state-of-the-art deep learning methods.

Suggestion mining from online reviews using temporal convolutional network

Business and brand owners are using social media networks to provide and deliver various services to their clients and collect information about their products from customers. Customers give their opinions as well as ideas for the improvement of the products on the review platforms and portals. Suggestion Mining is a technique of automatic extraction of these innovative ideas or suggestions from online source data. In this paper, we proposed TCN architecture for suggestion mining from online reviews. The TCN uses causal and dilated convolutional layers to process sequential or temporal data and captures long-term dependencies. TCN architecture on the dataset of SemEval-2019 subtask A is experimented. The dataset is highly imbalanced and to overcome this problem, the ensemble oversampling technique to balance the dataset is applied. TCN is also experimented with the attention mechanism. Our proposed model outperforms the existing works by achieving an F1 score of 82.0 %.

Dynamic load identification based on deep convolution neural network

The deep learning methods have been extensively studied in the field of dynamic load identification, due to their strong direct modeling ability between vibration response and external excitation. Dynamic load identification is a complicated inverse problem, which extremely relies on the solution of the structural model parameter. Nevertheless, the accurate computation of model parameters is always a challenge, small errors in model parameters will lead to inaccuracy of dynamic load identification. This brings various hardships to engineering applications. To achieve this problem, we propose a novel method based on a deep dilated convolution neural network (DCNN) for dynamic load identification, directly constructing the inverse model between vibration response and excitation, avoiding solving the model parameter. A dynamic load identification model, which contains two 1-D dilated convolution layers, one pooling layer, and two fully connected layers, is constructed to estimate the sinusoidal, impact, and random dynamic loads of a simply supported beam. We also appraise the anti-noise ability of the proposed method for load identification. Moreover, a vibration test is carried out to further evaluate this algorithm in the experimental aspect. Additionally, we analyze the comparison of the proposed method along with the Green kernel function method, and the dynamic system with uncertain model parameters is also analyzed. Besides, the operation of the convolution layer for response input is studied, and the applicability to different distributions of measurement points and the adaptability for frequency domain data are investigated. Ultimately, we find this method has a strong anti-noise ability due to its convolution layer, which can be regarded as a filter in dynamic load identification. Furthermore, the proposed method is of great practical for engineering applications owing to its satisfying applicability for systems with uncertain parameters, distributions of measurement points, and frequency data. All the results reveal the advantages of the identification method based on DCNN, consisting of good accuracy, reliability, and robustness. These results can be favorable for many applications.

SC-CAN: Spectral Convolution and Channel Attention Network for Wheat Stress Classification

Biotic and abiotic plant stress (e.g., frost, fungi, diseases) can significantly impact crop production. It is thus essential to detect such stress at an early stage before visual symptoms and damage become apparent. To this end, this paper proposes a novel deep learning method, called Spectral Convolution and Channel Attention Network (SC-CAN), which exploits the difference in spectral responses of healthy and stressed crops. The proposed SC-CAN method comprises two main modules: (i) a spectral convolution module, which consists of dilated causal convolutional layers stacked in a residual manner to capture the spectral features; (ii) a channel attention module, which consists of a global pooling layer and fully connected layers that compute inter-relationship between feature map channels before scaling them based on their importance level (attention score). Unlike standard convolution, which focuses on learning local features, the dilated convolution layers can learn both local and global features. These layers also have long receptive fields, making them suitable for capturing long dependency patterns in hyperspectral data. However, because not all feature maps produced by the dilated convolutional layers are important, we propose a channel attention module that weights the feature maps according to their importance level. We used SC-CAN to classify salt stress (i.e., abiotic stress) on four datasets (Chinese Spring (CS), Aegilops columnaris (co(CS)), Ae. speltoides auchery (sp(CS)), and Kharchia datasets) and Fusarium head blight disease (i.e., biotic stress) on Fusarium dataset. Reported experimental results show that the proposed method outperforms existing state-of-the-art techniques with an overall accuracy of 83.08%, 88.90%, 82.44%, 82.10%, and 82.78% on CS, co(CS), sp(CS), Kharchia, and Fusarium datasets, respectively.

Temporal Context Modeling Network with Local-Global Complementary Architecture for Temporal Proposal Generation

Temporal Action Proposal Generation (TAPG) is a promising but challenging task with a wide range of practical applications. Although state-of-the-art methods have made significant progress in TAPG, most ignore the impact of the temporal scales of action and lack the exploitation of effective boundary contexts. In this paper, we propose a simple but effective unified framework named Temporal Context Modeling Network (TCMNet) that generates temporal action proposals. TCMNet innovatively uses convolutional filters with different dilation rates to address the temporal scale issue. Specifically, TCMNet contains a BaseNet with dilated convolutions (DBNet), an Action Completeness Module (ACM), and a Temporal Boundary Generator (TBG). The DBNet aims to model temporal information. It handles input video features through different dilated convolutional layers and outputs a feature sequence as the input of ACM and TBG. The ACM aims to evaluate the confidence scores of densely distributed proposals. The TBG is designed to enrich the boundary context of an action instance. The TBG can generate action boundaries with high precision and high recall through a local–global complementary structure. We conduct comprehensive evaluations on two challenging video benchmarks: ActivityNet-1.3 and THUMOS14. Extensive experiments demonstrate the effectiveness of the proposed TCMNet on tasks of temporal action proposal generation and temporal action detection.

Application of Artificial Neural Network in College-Level Music Teaching Quality Evaluation

Art is a very practical activity, especially music art. Quality music education at colleges and universities is vital for the ideological and moral schooling of students. Owing to the rapid expansion of music education at graduate level, the building of therapeutic evaluation of music instruction is critical. In fact, most of the music instruction in colleges and universities has not built a scientific and appropriate evaluation system based on the actual teaching quality of the classroom. This work combines the emerging neural network (NN) technology with standard method of music teaching evaluation and proposes a novel method—Music Teaching Quality Evaluation Network (MTQEN). To effectively improve performance of the model, the method uses a one-dimensional convolutional neural network (1D-CNN) being optimized from three aspects: expanding the receptive field, reducing training parameters, and enhancing operationality. Instead of the traditional convolution layer, the dilated convolution layer is utilized to increase scope of local receptive field and to improve the feature extraction efficacy. To improve training rate and to eliminate dependency on batch size, the filter response normalization (FRN) is used. Moreover, global pooling is used to reduce the requisite training parameters and to improve the training efficiency. Results of the evaluation show that performance of MTQEN in term accuracy (95.6%) and recall (93.3%) is better than the other contemporary models. The method proposed has great significance in pedagogy of music and arts, whereas the network designed may be enhanced to effectively evaluate teaching in related domains as well.

An Improved YOLO Algorithm for Fast and Accurate Underwater Object Detection

Due to the abundant natural resources of the underwater world, autonomous exploration using underwater robots has become an effective technological tool in recent years. Real-time object detection is critical when employing robots for independent underwater exploration. However, when a robot detects underwater, its computing power is usually limited, which makes it challenging to detect objects effectively. To solve this problem, this study presents a novel algorithm for underwater object detection based on YOLOv4-tiny to achieve better performance with less computational cost. First, a symmetrical bottleneck-type structure is introduced into the YOLOv4-tiny’s backbone network based on dilated convolution and 1 × 1 convolution. It captures contextual information in feature maps with reasonable computational cost and improves the mAP score by 8.74% compared to YOLOv4-tiny. Second, inspired by the convolutional block attention module, a symmetric FPN-Attention module is constructed by integrating the channel-attention module and the spatial-attention module. Features extracted by the backbone network can be fused more efficiently by the symmetric FPN-Attention module, achieving a performance improvement of 8.75% as measured by mAP score compared to YOLOv4-tiny. Finally, this work proposed the YOLO-UOD for underwater object detection through the fusion of the YOLOv4-tiny structure, symmetric FPN-Attention module, symmetric bottleneck-type dilated convolutional layers, and label smoothing training strategy. It can efficiently detect underwater objects in an embedded system environment with limited computing power. Experiments show that the proposed YOLO-UOD outperforms the baseline model on the Brackish underwater dataset, with a detection mAP of 87.88%, 10.5% higher than that of YOLOv4-tiny’s 77.38%, and the detection result exceeds YOLOv5s’s 83.05% and YOLOv5m’s 84.34%. YOLO-UOD is deployed on the embedded system Jetson Nano 2 GB with a detection speed of 9.24 FPS, which shows that it can detect effectively in scenarios with limited computing power.

A New Approach to Quantify and Grade Radiation Dermatitis Using Deep-Learning Segmentation in Skin Photographs

AimsObjective evaluation of radiation dermatitis is important for analysing the correlation between the severity of radiation dermatitis and dose distribution in clinical practice and for reliable reporting in clinical trials. We developed a novel radiation dermatitis segmentation system based on convolutional neural networks (CNNs) to consistently evaluate radiation dermatitis. Materials and methodsThe radiation dermatitis segmentation system is designed to segment the radiation dermatitis occurrence area using skin photographs and skin-dose distribution. A CNN architecture with a dilated convolution layer and skip connection was designed to estimate the radiation dermatitis area. Seventy-three skin photographs obtained from patients undergoing radiotherapy were collected for training and testing. The ground truth of radiation dermatitis segmentation is manually delineated from the skin photograph by an experienced radiation oncologist and medical physicist. We converted the skin photographs to RGB (red-green-blue) and CIELAB (lightness (L∗), red-green (a∗) and blue-yellow (b∗)) colour information and trained the network to segment faint and severe radiation dermatitis using three different input combinations: RGB, RGB + CIELAB (RGBLAB) and RGB + CIELAB + skin-dose distribution (RGBLAB_D). The proposed system was evaluated using the Dice similarity coefficient (DSC), sensitivity, specificity and normalised Matthews correlation coefficient (nMCC). A paired t-test was used to compare the results of different segmentation performances. ResultsOptimal data composition was observed in the network trained for radiation dermatitis segmentation using skin photographs and skin-dose distribution. The average DSC, sensitivity, specificity and nMCC values of RGBLAB_D were 0.62, 0.61, 0.91 and 0.77, respectively, in faint radiation dermatitis, and 0.69, 0.78, 0.96 and 0.83, respectively, in severe radiation dermatitis. ConclusionOur study showed that CNN-based radiation dermatitis segmentation in skin photographs of patients undergoing radiotherapy can describe radiation dermatitis severity and pattern. Our study could aid in objectifying the radiation dermatitis grading and analysing the reliable correlation between dosimetric factors and the morphology of radiation dermatitis.

Crowd counting in complex scenes based on an attention aware CNN network

Crowd counting with density estimation has been an active research community due to its significant applications in the fields of public security, video surveillance, traffic monitoring. However, Crowd counting for congested scenes often suffers from some obstacles including severe occlusions, large scale variations, noise interference, etc. In this paper, using the first ten layers of a modified VGG16 and dilated convolution layers as the framework, we have proposed a CNN based crowd counting and density estimation model improved by the attention aware modules with residual connections. To tackle the problem of noise interference, convolutional block attention modules have been introduced into the deep network to segment the foreground and background to focus on interest information, refining deeper features of the input image. To improve information transmission and reuse, residual connections are utilized to link 3 attention blocks. Meanwhile, dilated convolution layers keep larger reception fields and obtain high-resolution density maps. The proposed method has been evaluated on three public benchmarks, i.e. Shanghai Tech A & B, UCF-QNRF and MALL, achieving the mean absolute errors of 64.6 & 8.3, 113.8 and 1.68, respectively. The results outperform some existing excellent approaches. This indicates that the proposed model has high accuracy and better robustness, which is suitable for crowd counting and density estimation in various congested scenes.

Unified DeepLabV3+ for Semi-Dark Image Semantic Segmentation

Semantic segmentation for accurate visual perception is a critical task in computer vision. In principle, the automatic classification of dynamic visual scenes using predefined object classes remains unresolved. The challenging problems of learning deep convolution neural networks, specifically ResNet-based DeepLabV3+ (the most recent version), are threefold. The problems arise due to (1) biased centric exploitations of filter masks, (2) lower representational power of residual networks due to identity shortcuts, and (3) a loss of spatial relationship by using per-pixel primitives. To solve these problems, we present a proficient approach based on DeepLabV3+, along with an added evaluation metric, namely, Unified DeepLabV3+ and , respectively. The presented unified version reduced the effect of biased exploitations via additional dilated convolution layers with customized dilation rates. We further tackled the problem of representational power by introducing non-linear group normalization shortcuts to solve the focused problem of semi-dark images. Meanwhile, to keep track of the spatial relationships in terms of the global and local contexts, geometrically bunched pixel cues were used. We accumulated all the proposed variants of DeepLabV3+ to propose Unified DeepLabV3+ for accurate visual decisions. Finally, the proposed evaluation metric was based on the weighted combination of three different accuracy measures, i.e., the pixel accuracy, IoU (intersection over union), and Mean BFScore, as robust identification criteria. Extensive experimental analysis performed over a CamVid dataset confirmed the applicability of the proposed solution for autonomous vehicles and robotics for outdoor settings. The experimental analysis showed that the proposed Unified DeepLabV3+ outperformed DeepLabV3+ by a margin of 3% in terms of the class-wise pixel accuracy, along with a higher , depicting the effectiveness of the proposed approach.

DSCNN: Dilated Shuffle CNN Model for SSVEP Signal Classification

The steady state visual evoked potential (SSVEP) based brain-computer interface (BCI) system has attracted a lot of attention. It is a great challenge to increase the classification accuracy of SSVEP, especially in fatigue state. In this paper, we propose a dilated shuffle convolutional neural network (DSCNN) model to realize EEG-based SSVEP signal classification. Firstly, we conduct experiments to obtain SSVEP recordings in normal and fatigue states. Then combining continuous wavelet transform (CWT) and DSCNN, we construct a framework for realizing the SSVEP detection. In DSCNN, the signals are processed by three parallel dilated convolution layers firstly, then we extract the characteristics of the signals through channel shuffle and group convolution, while reducing the computational load. For normal condition, we reach an average accuracy rate of 96.75%, and for the data under fatigue state, the average accuracy of this method increases to 77.52%. Through the comparison with the existing methods, the effectiveness and advance of our method are proved, and the effect of channel shuffle on signal extraction is also demonstrated by comparison.

Precise Crop Classification of Hyperspectral Images Using Multi-Branch Feature Fusion and Dilation-Based MLP

The precise classification of crop types using hyperspectral remote sensing imaging is an essential application in the field of agriculture, and is of significance for crop yield estimation and growth monitoring. Among the deep learning methods, Convolutional Neural Networks (CNNs) are the premier model for hyperspectral image (HSI) classification for their outstanding locally contextual modeling capability, which facilitates spatial and spectral feature extraction. Nevertheless, the existing CNNs have a fixed shape and are limited to observing restricted receptive fields, constituting a simulation difficulty for modeling long-range dependencies. To tackle this challenge, this paper proposed two novel classification frameworks which are both built from multilayer perceptrons (MLPs). Firstly, we put forward a dilation-based MLP (DMLP) model, in which the dilated convolutional layer replaced the ordinary convolution of MLP, enlarging the receptive field without losing resolution and keeping the relative spatial position of pixels unchanged. Secondly, the paper proposes multi-branch residual blocks and DMLP concerning performance feature fusion after principal component analysis (PCA), called DMLPFFN, which makes full use of the multi-level feature information of the HSI. The proposed approaches are carried out on two widely used hyperspectral datasets: Salinas and KSC; and two practical crop hyperspectral datasets: WHU-Hi-LongKou and WHU-Hi-HanChuan. Experimental results show that the proposed methods outshine several state-of-the-art methods, outperforming CNN by 6.81%, 12.45%, 4.38% and 8.84%, and outperforming ResNet by 4.48%, 7.74%, 3.53% and 6.39% on the Salinas, KSC, WHU-Hi-LongKou and WHU-Hi-HanChuan datasets, respectively. As a result of this study, it was confirmed that the proposed methods offer remarkable performances for hyperspectral precise crop classification.

A Dual Discriminator Adversarial Learning Approach for Dental Occlusal Surface Reconstruction.

Objective Restoring the correct masticatory function of partially edentulous patient is a challenging task primarily due to the complex tooth morphology between individuals. Although some deep learning-based approaches have been proposed for dental restorations, most of them do not consider the influence of dental biological characteristics for the occlusal surface reconstruction. Description. In this article, we propose a novel dual discriminator adversarial learning network to address these challenges. In particular, this network architecture integrates two models: a dilated convolutional-based generative model and a dual global-local discriminative model. While the generative model adopts dilated convolution layers to generate a feature representation that preserves clear tissue structure, the dual discriminative model makes use of two discriminators to jointly distinguish whether the input is real or fake. While the global discriminator focuses on the missing teeth and adjacent teeth to assess whether it is coherent as a whole, the local discriminator aims only at the defective teeth to ensure the local consistency of the generated dental crown. Results Experiments on 1000 real-world patient dental samples demonstrate the effectiveness of our method. For quantitative comparison, the image quality metrics are used to measure the similarity of the generated occlusal surface, and the root mean square between the generated result and the target crown calculated by our method is 0.114 mm. In qualitative analysis, the proposed approach can generate more reasonable dental biological morphology. Conclusion The results demonstrate that our method significantly outperforms the state-of-the-art methods in occlusal surface reconstruction. Importantly, the designed occlusal surface has enough anatomical morphology of natural teeth and superior clinical application value.

An effective modular approach for crowd counting in an image using convolutional neural networks

Abrupt and continuous nature of scale variation in a crowded scene is a challenging task to enhance crowd counting accuracy in an image. Existing crowd counting techniques generally used multi-column or single-column dilated convolution to tackle scale variation due to perspective distortion. However, due to multi-column nature, they obtain identical features, whereas, the standard dilated convolution (SDC) with expanded receptive field size has sparse pixel sampling rate. Due to sparse nature of SDC, it is highly challenging to obtain relevant contextual information. Further, features at multiple scale are not extracted despite some inception-based model is not used (which is cost effective). To mitigate theses drawbacks in SDC, we therefore, propose a hierarchical dense dilated deep pyramid feature extraction through convolution neural network (CNN) for single image crowd counting (HDPF). It comprises of three modules: general feature extraction module (GFEM), deep pyramid feature extraction module (PFEM) and fusion module (FM). The GFEM is responsible to obtain task independent general features. Whereas, PFEM plays a vital role to obtain the relevant contextual information due to dense pixel sampling rate caused by densely connected dense stacked dilated convolutional modules (DSDCs). Further, due to dense connections among DSDCs, the final feature map acquires multi-scale information with expanded receptive field as compared to SDC. Due to dense pyramid nature, it is very effective to propagate the extracted feature from lower dilated convolutional layers (DCLs) to middle and higher DCLs, which result in better estimation accuracy. The FM is used to fuse the incoming features extracted by other modules. The proposed technique is tested through simulations on three well known datasets: Shanghaitech (Part-A), Shanghaitech (Part-B) and Venice. Results justify its relative effectiveness in terms of selected performance.

Deep learning characterization of surface defects in the selective laser melting process

Surface defects in the selective laser melting (SLM) process adversely affect the surface quality of additive manufacturing workpieces. Although some studies have been conducted to classify or detect defects, the investigation of the distribution of defects has received relatively little attention. Currently, there are many studies which focus on the counting of objects and they typically adopt a backbone convolutional neural network to obtain an initial feature map, which contains more semantic information but loses some geometric details. In this paper, the distribution of surface defects and defect count estimation of additive manufactured components are first studied as well as presentation of a developed deep learning characterization method (DLCM) based on a detail-aware dilated convolutional neural network (DDCNN) incorporated with a fine details feature map extractor designed to obtain final fine semantic features. It features two dilated convolutional layer combination blocks (DCBs), which are proposed to fuse low-level features with semantic features. Data are acquired from the surface of a workpiece fabricated by SLM. A series of experiments have been conducted to validate the performance of the DLCM. Compared with the other main state-of-the-art methods, the proposed DLCM yields better results.

Skeleton-Based Spatio-Temporal U-Network for 3D Human Pose Estimation in Video.

Despite the great progress in 3D pose estimation from videos, there is still a lack of effective means to extract spatio-temporal features of different granularity from complex dynamic skeleton sequences. To tackle this problem, we propose a novel, skeleton-based spatio-temporal U-Net(STUNet) scheme to deal with spatio-temporal features in multiple scales for 3D human pose estimation in video. The proposed STUNet architecture consists of a cascade structure of semantic graph convolution layers and structural temporal dilated convolution layers, progressively extracting and fusing the spatio-temporal semantic features from fine-grained to coarse-grained. This U-shaped network achieves scale compression and feature squeezing by downscaling and upscaling, while abstracting multi-resolution spatio-temporal dependencies through skip connections. Experiments demonstrate that our model effectively captures comprehensive spatio-temporal features in multiple scales and achieves substantial improvements over mainstream methods on real-world datasets.