Search

Existing lightweight video super-resolution network architectures are often simple in structure and lack generalization ability when dealing with complex and varied real scenes in aerial videos of unmanned aerial vehicle. Furthermore, these networks may cause issues such as the checkerboard effect and loss of texture information when processing drone videos. To address these challenges, we propose a super-lightweight video super-resolution reconstruction network based on convolutional pyramids and progressive residual blocks: ULSR-UV. The ULSR-UV network significantly reduces model redundancy and achieves high levels of lightness by incorporating a 3D lightweight spatial pyramid structure and more efficient residual block designs. This network utilizes a specific optimizer to efficiently process drone videos from both multi-frame and single-frame dimensions. Additionally, the ULSR-UV network incorporates a multidimensional feature loss calculation module that enhances network performance and significantly improves the reconstruction quality of drone aerial videos. Extensive experimental verification has demonstrated ULSR-UV's outstanding performance in the field of drone video super-resolution reconstruction. [ABSTRACT FROM AUTHOR]

Deep learning models have been proven to be susceptible to malicious adversarial attacks, which manipulate input images to deceive the model into making erroneous decisions. Consequently, the threat posed to these models serves as a poignant reminder of the necessity to focus on the model security of object segmentation algorithms based on deep learning. However, the current landscape of research on adversarial attacks primarily centers around static images, resulting in a dearth of studies on adversarial attacks targeting Video Object Segmentation (VOS) models. Given that a majority of self-supervised VOS models rely on affinity matrices to learn feature representations of video sequences and achieve robust pixel correspondence, our investigation has delved into the impact of adversarial attacks on self-supervised VOS models. In response, we propose an innovative black-box attack method incorporating contrastive loss. This method induces segmentation errors in the model through perturbations in the feature space and the application of a pixel-level loss function. Diverging from conventional gradient-based attack techniques, we adopt an iterative black-box attack strategy that incorporates contrastive loss across the current frame, any two consecutive frames, and multiple frames. Through extensive experimentation conducted on the DAVIS 2016 and DAVIS 2017 datasets using three selfsupervised VOS models and one unsupervised VOS model, we unequivocally demonstrate the potent attack efficiency of the black-box approach. Remarkably, the J&F metric value experiences a significant decline of up to 50.08% post-attack. [ABSTRACT FROM AUTHOR]

The extraction of photovoltaic (PV) panels from remote sensing images is of great significance for estimating the power generation of solar photovoltaic systems and informing government decisions. The implementation of existing methods often struggles with complex background interference and confusion between the background and the PV panels. As a result, the completeness and edge clarity of PV panel extraction results are compromised. Moreover, most previous studies have overlooked the unique color characteristics of PV panels. To alleviate these deficiencies and limitations, a method for extracting photovoltaic panels from high-resolution optical remote sensing images guided by prior knowledge (PKGPVN) is proposed. Firstly, aiming to address the problems related to missed extractions and background misjudgments, a Photovoltaic Index (PVI) based on visible images in the three-band is constructed to serve as prior knowledge to differentiate between PV panels and non-PV panels. Secondly, in order to strengthen information interaction between shallow features and deep features and enhance the accuracy and integrity of results, a Residual Convolution Hybrid Attention Module (RCHAM) is introduced into the skip-connection of the encoding–decoding structure. Finally, for the purpose of reducing the phenomenon of blurred edges, a multilevel Feature Loss (FL) function is designed to monitor the prediction results at different scales. Comparative experiments are conducted with seven methods, including U-Net, on publicly available datasets. The experimental results show that our PKGPVN achieves superior performance in terms of evaluation metrics such as IoU (above 82%), Precision (above 91%), Recall (above 89%), and F1-score (above 90%) on the AIR-PV dataset. Additionally, the ablation experiments illustrate the effectiveness of our key parts. The proposed method reduces the phenomena of missed extractions and background misjudgments effectively while producing highly accurate results with clear boundaries. [ABSTRACT FROM AUTHOR]

Phylogenetic metrics are essential tools used in the study of ecology, evolution and conservation. Phylogenetic diversity (PD) in particular is one of the most prominent measures of biodiversity and is based on the idea that biological features accumulate along the edges of phylogenetic trees that are summed. We argue that PD and many other phylogenetic biodiversity metrics fail to capture an essential process that we term attrition. Attrition is the gradual loss of features through causes other than extinction. Here we introduce "EvoHeritage", a generalization of PD that is founded on the joint processes of accumulation and attrition of features. We argue that while PD measures evolutionary history, EvoHeritage is required to capture a more pertinent subset of evolutionary history including only components that have survived attrition. We show that EvoHeritage is not the same as PD on a tree with scaled edges; instead, accumulation and attrition interact in a more complex non-monophyletic way that cannot be captured by edge lengths alone. This leads us to speculate that the one-dimensional edge lengths of classic trees may be insufficiently flexible to capture the nuances of evolutionary processes. We derive a measure of EvoHeritage and show that it elegantly reproduces species richness and PD at opposite ends of a continuum based on the intensity of attrition. We demonstrate the utility of EvoHeritage in ecology as a predictor of community productivity compared with species richness and PD. We also show how EvoHeritage can quantify living fossils and resolve their associated controversy. We suggest how the existing calculus of PD-based metrics and other phylogenetic biodiversity metrics can and should be recast in terms of EvoHeritage accumulation and attrition. [ABSTRACT FROM AUTHOR]

Shale gas, as a kind of unconventional resources, plays an important role in meeting global energy demands and is expected to do so for decades as human society is moving toward low-carbon energies. During the energy transition, the rapid growth of shale gas helps to maintain the security of supply and lower prices of energies for consumers. The organic matters contained in shale are the physical basis for the production of shale gas, and the large number of pores and fractures in shale provide storage space for shale gas. Therefore, shale rocks are both source rocks for gas generation and reservoirs for gathering and preserving gas. Accurate characterization and modeling of the internal spatial structure of shale is of great significance for the exploitation of shale gas. However, as shale contains a variety of minerals and organic matters, and the scales of pores and fractures vary widely, using conventional numerical methods to characterize and reconstruct the internal structures of shale is challenging. When reconstructing some larger results with high resolutions, the contradiction between the limited field of view (FOV) and the resolution of training images becomes more acute. With the development of deep learning and one of its variants, generative adversarial network (GAN), the reconstruction of shale may benefit from their strong learning and generation abilities. In this paper, a model of GAN combined with residual networks (3DRGAN) is proposed to reconstruct 3D digital cores of shale. The super-resolution image that is much bigger than training images is obtained, solving the contradiction between FOV and the resolution of training images. Compared to some typical methods, the results reconstructed by 3DRGAN are closer to the real shale sample, showing its favorable advantage. [ABSTRACT FROM AUTHOR]

Featured Application: Increased parallelism in edge network security issues, Feature loss handling. Feature loss in IoT scenarios is a common problem. This situation poses a greater challenge in terms of real-time and accuracy for the security of intelligent edge computing systems, which also includes network security intrusion detection systems (NIDS). Losing some packet information can easily confuse NIDS and cause an oversight of security systems. We propose a novel network intrusion detection framework based on an improved neural network. The new framework uses 23 subframes and a mixer for multi-classification work, which improves the parallelism of NIDS and is more adaptable to edge networks. We also incorporate the K-Nearest Neighbors (KNN) algorithm and Genetic Algorithm (GA) for feature selection, reducing parameters, communication, and memory overhead. We named the above system as Combinatorial Multi-Classification-NIDS (CM-NIDS). Experiments demonstrate that our framework can be more flexible in terms of the parameters of binary classification, has a fairly high accuracy in multi-classification, and is less affected by feature loss. [ABSTRACT FROM AUTHOR]

Although medical imaging is frequently used to diagnose diseases, in complex diagnostic situations, specialists typically need to look at different modalities of image information. Creating a composite multimodal medical image can aid professionals in making quick and accurate diagnoses of diseases. The fused images of many medical image fusion algorithms, however, are frequently unable to precisely retain the functional and structural information of the source image. This work develops an end-to-end model based on GAN (U-Patch GAN) to implement the self-supervised fusion of multimodal brain images in order to enhance the fusion quality. The model uses the classical network U-net as the generator, and it uses the dual adversarial mechanism based on the Markovian discriminator (PatchGAN) to enhance the generator's attention to high-frequency information. To ensure that the network satisfies the Lipschitz continuity, we apply the spectral norm to each layer of the network. We also propose better adversarial loss and feature loss (feature matching loss and VGG-16 perceptual loss) based on the F-norm, which significantly enhance the quality of fused images. On public data sets, we performed a lot of tests. First, we studied how clinically useful the fused image was. The model's performance in single-slice images and continuous-slice images was then confirmed by comparison with other six most popular mainstream fusion approaches. Finally, we verify the effectiveness of the adversarial loss and feature loss. [ABSTRACT FROM AUTHOR]

The extraction of photovoltaic (PV) panels from remote sensing images is of great significance for estimating the power generation of solar photovoltaic systems and informing government decisions. The implementation of existing methods often struggles with complex background interference and confusion between the background and the PV panels. As a result, the completeness and edge clarity of PV panel extraction results are compromised. Moreover, most previous studies have overlooked the unique color characteristics of PV panels. To alleviate these deficiencies and limitations, a method for extracting photovoltaic panels from high-resolution optical remote sensing images guided by prior knowledge (PKGPVN) is proposed. Firstly, aiming to address the problems related to missed extractions and background misjudgments, a Photovoltaic Index (PVI) based on visible images in the three-band is constructed to serve as prior knowledge to differentiate between PV panels and non-PV panels. Secondly, in order to strengthen information interaction between shallow features and deep features and enhance the accuracy and integrity of results, a Residual Convolution Hybrid Attention Module (RCHAM) is introduced into the skip-connection of the encoding–decoding structure. Finally, for the purpose of reducing the phenomenon of blurred edges, a multilevel Feature Loss (FL) function is designed to monitor the prediction results at different scales. Comparative experiments are conducted with seven methods, including U-Net, on publicly available datasets. The experimental results show that our PKGPVN achieves superior performance in terms of evaluation metrics such as IoU (above 82%), Precision (above 91%), Recall (above 89%), and F1-score (above 90%) on the AIR-PV dataset. Additionally, the ablation experiments illustrate the effectiveness of our key parts. The proposed method reduces the phenomena of missed extractions and background misjudgments effectively while producing highly accurate results with clear boundaries.

Aiming at several problems of convolutional neural network(CNN) model such as the amount of parameters to be trained is too large and the multi-layer network leads to the disappearance of features and gradients, a CNN model based on split layers and Inception- ResNet for intrusion detection is proposed. First, to make it suitable for the input of the CNN, the network traffic data was converted into spatially correlated images. Then the Inception-ResNet convolutional layer was introduced to increase the depth of the network, reduce the amount of training parameters and eliminate the problem of gradient disappearance. The split layer basically maintained the original classification features of the image when reducing the image scale. Finally, the NSL-KDD data set was used to train and test the improved model. The test results show that compared with the Inception-ResNet model, the improved model has higher detection accuracy rate, higher recall rate of minority class and better training time performance. [ABSTRACT FROM AUTHOR]

Due to the images obtained in lowlight environments often showing low contrast, low brightness and artifacts, it is difficult to distinguish the details of these images for people. In the field of images fusion and target tacking, lowlight images cannot be processed better. In this paper, we proposed an end-to-end lowlight image enhancement network, which uses modules stacking methods and attention modules. Firstly, the method of module stacking was applied to extract the different features of images, and then the features are fused on the channel dimension. Finally, the final image was reconstructed with a series of convolutions. In particular, our loss function consists of two parts: the first part of the loss function was calculated using L1 loss, L2 loss and the gradient loss, and VGG network was utilized to calculate the second part. Furthermore, we verified the effectiveness of the model via a large number of comparative experiments, and illustrated the comparison results through quantitative and qualitative methods. We additionally show the performance of our network on lowlight video enhancement, which also has better results than the other methods. [ABSTRACT FROM AUTHOR]

Existing video inpainting methods usually produce blurred texture,distorted structure and artifacts,while applying image-based inpainting model directly to the video inpainting will lead to inconsistent time.From the perspective of time,a novel coherent semantic spatial-temporal attention(CSSTA) for video inpainting is proposed,through the attention layer,the model focuses on the information that the target frame is partially blocked and the adjacent frames are visible,so as to obtain the visible content to fill the hole region of the target frame.The CSSTA layer can not only model the semantic correlation between hole features but also remotely correlate the long-range information with the hole regions.In order to complete semantically coherent hole regions,a novel loss function Feature Loss is proposed to replace VGG Loss.The model is built on a two-stage coarse-to-fine encoder-decoder model for collecting and refining information from adjacent frames.Experimental results on the YouTube-VOS and DAVIS datasets show that the method in this paper runs almost in real-time and outperforms the three typical video inpainting methods in terms of inpainting results,peak signal-to-noise ratio (PSNR) and structural similarity (SSIM).

Feature loss in IoT scenarios is a common problem. This situation poses a greater challenge in terms of real-time and accuracy for the security of intelligent edge computing systems, which also includes network security intrusion detection systems (NIDS). Losing some packet information can easily confuse NIDS and cause an oversight of security systems. We propose a novel network intrusion detection framework based on an improved neural network. The new framework uses 23 subframes and a mixer for multi-classification work, which improves the parallelism of NIDS and is more adaptable to edge networks. We also incorporate the K-Nearest Neighbors (KNN) algorithm and Genetic Algorithm (GA) for feature selection, reducing parameters, communication, and memory overhead. We named the above system as Combinatorial Multi-Classification-NIDS (CM-NIDS). Experiments demonstrate that our framework can be more flexible in terms of the parameters of binary classification, has a fairly high accuracy in multi-classification, and is less affected by feature loss.

Purpose: To improve reconstruction fidelity of fine structures and textures in deep learning‐ (DL) based reconstructions. Methods: A novel patch‐based Unsupervised Feature Loss (UFLoss) is proposed and incorporated into the training of DL‐based reconstruction frameworks in order to preserve perceptual similarity and high‐order statistics. The UFLoss provides instance‐level discrimination by mapping similar instances to similar low‐dimensional feature vectors and is trained without any human annotation. By adding an additional loss function on the low‐dimensional feature space during training, the reconstruction frameworks from under‐sampled or corrupted data can reproduce more realistic images that are closer to the original with finer textures, sharper edges, and improved overall image quality. The performance of the proposed UFLoss is demonstrated on unrolled networks for accelerated two‐ (2D) and three‐dimensional (3D) knee MRI reconstruction with retrospective under‐sampling. Quantitative metrics including normalized root mean squared error (NRMSE), structural similarity index (SSIM), and our proposed UFLoss were used to evaluate the performance of the proposed method and compare it with others. Results: In vivo experiments indicate that adding the UFLoss encourages sharper edges and more faithful contrasts compared to traditional and learning‐based methods with pure ℓ2$$ {\ell}_2 $$ loss. More detailed textures can be seen in both 2D and 3D knee MR images. Quantitative results indicate that reconstruction with UFLoss can provide comparable NRMSE and a higher SSIM while achieving a much lower UFLoss value. Conclusion: We present UFLoss, a patch‐based unsupervised learned feature loss, which allows the training of DL‐based reconstruction to obtain more detailed texture, finer features, and sharper edges with higher overall image quality under DL‐based reconstruction frameworks. (Code available at: https://github.com/mikgroup/UFLoss) [ABSTRACT FROM AUTHOR]

Methods based on generative adversarial network (GAN) have been widely used in infrared and visible images fusion. However, these methods cannot perceive the discriminative parts of an image. Therefore, we introduce a multigrained attention module into encoder–decoder network to fuse infrared and visible images (MgAN-Fuse). The infrared and visible images are encoded by two independent encoder networks due to their diverse modalities. Then, the results of the two encoders are concatenated to calculate the fused result by the decoder. To exploit the features of multiscale layers fully and force the model focus on the discriminative regions, we integrate attention modules into multiscale layers of the encoder to obtain multigrained attention maps, and then, the multigrained attention maps are concatenated with the corresponding multiscale features of the decoder network. Thus, the proposed method can preserve the foreground target information of the infrared image and capture the context information of the visible image. Furthermore, we design an additional feature loss in the training process to preserve the important features of the visible image, and a dual adversarial architecture is employed to help the model capture enough infrared intensity information and visible details simultaneously. The ablation studies illustrate the validity of the multigrained attention network and feature loss function. Extensive experiments on two infrared and visible image data sets demonstrate that the proposed MgAN-Fuse has a better performance than state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Target detection plays a core issue in hyperspectral remote sensing, but faces serious challenges of how to deal with the spatial and spectral redundancies and spectral variations. In this article, a novel network block is developed, called RFT-MSA block (abbreviated as RM), which includes the region-of-interest (RoI) feature transformation (RFT) and the multiscale-spectral-attention (MSA) module as to reduce the spatial and spectral redundancies simultaneously and provide strong discrimination. Furthermore, a deep spatial–spectral network (DSSN) is presented by stacking several RM and deconvolutional (DC) blocks for hyperspectral target detection in an unsupervised manner, and a feature loss term is investigated to simultaneously restrict the target to be sparse and minimize the energy of the background. The proposed algorithm mainly consists of three steps. First, an RoI map is detected using a classical detector (no statistic assumption is needed) with an edge-preserving filter. Then, the hyperspectral image (HSI) and the corresponding RoI map are considered as inputs to the DSSN for extracting the spatial and spectral feature of interest (SSFI). Finally, we apply the nearest neighbors (NNs) to the SSFI for detection-map refinement. The experimental results on one synthetic and three real HSIs demonstrate that the proposed algorithm outperforms other benchmark approaches in detection performance and robustness. In addition, further analysis also demonstrates the effectiveness of the proposed RM block. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Jilin University (Science Edition) / Jilin Daxue Xuebao (Lixue Ban) is the property of Zhongguo Xue shu qi Kan (Guang Pan Ban) Dian zi Za zhi She and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

This paper presents an advanced methodology that combines Convolutional Neural Networks (CNNs) and Transformers to tackle the issue of feature loss and elevate segmentation accuracy. While Transformers excel at capturing global features, there is a potential risk of losing crucial global information when CNNs concentrate on local feature extraction. To counteract this, we introduce the FM module, which re-extracts local information from the Encoder's output, reinforcing local feature expression and enhancing segmentation accuracy. Additionally, our approach places a significant emphasis on the judicious fusion of features in the Decoder. By incorporating a technology that aggregates both global and local features, we aim to prevent the loss of feature information. To augment the effective representation of data features and heighten accuracy across diverse segmentation tasks, we introduce the GS Feature Combination method, which adjusts the weights of different features during the aggregation process. Experimental results demonstrate noteworthy performance improvements, achieving an Intersection over Union (IOU) of 71.3%, surpassing existing methods. This innovative approach bears substantial importance in the diagnosis, treatment, and prognosis prediction of osteosarcoma, affording doctors the opportunity to reduce workload and time while upholding diagnostic precision. • Pathological image segmentation, crucial for medical diagnosis. • U-Net and others address challenges but risk losing essential features. • GSFormer integrates a transformer for global features and a CNN for localization. • High accuracy in analyzing osteosarcoma images. • GSFormer addresses medical deficiencies in developing countries. [ABSTRACT FROM AUTHOR]