Large Target Research Articles

Large-Kernel Central Block Masked Convolution and Channel Attention-Based Reconstruction Network for Anomaly Detection of High-Resolution Hyperspectral Imagery

In recent years, the rapid advancement of drone technology has led to an increasing use of drones equipped with hyperspectral sensors for ground imaging. Hyperspectral data captured via drones offer significantly higher spatial resolution, but this also introduces more complex background details and larger target scales in high-resolution hyperspectral imagery (HRHSI), posing substantial challenges for hyperspectral anomaly detection (HAD). Mainstream reconstruction-based deep learning methods predominantly emphasize spatial local information in hyperspectral images (HSIs), relying on small spatial neighborhoods for reconstruction. As a result, large anomalous targets and background details are often well reconstructed, leading to poor anomaly detection performance, as these targets are not sufficiently distinguished from the background. To address these limitations, we propose a novel HAD network for HRHSI based on large-kernel central block masked convolution and channel attention, termed LKCMCA. Specifically, we first employ the pixel-shuffle technique to reduce the size of anomalous targets without losing image information. Next, we design a large-kernel central block masked convolution to make the network pay more attention to the surrounding background information, enabling better fusion of the information between adjacent bands. This, coupled with an efficient channel attention mechanism, allows the network to capture deeper spectral features, enhancing the reconstruction of the background while suppressing anomalous targets. Furthermore, we introduce an adaptive loss function by down-weighting anomalous pixels based on the mean absolute error. This loss function is specifically designed to suppress the reconstruction of potentially anomalous pixels during network training, allowing our model to be considered an excellent background reconstruction network. By leveraging reconstruction error, the model effectively highlights anomalous targets. Meanwhile, we produced four benchmark datasets specifically for HAD tasks using existing HRHSI data, addressing the current shortage of HRHSI datasets in the HAD field. Extensive experiments demonstrate that our LKCMCA method achieves superior detection performance, outperforming ten state-of-the-art HAD methods on all datasets.

Hierarchical-Concatenate Fusion TDNN for sound event classification.

Semantic feature combination/parsing issue is one of the key problems in sound event classification for acoustic scene analysis, environmental sound monitoring, and urban soundscape analysis. The input audio signal in the acoustic scene classification is composed of multiple acoustic events, which usually leads to low recognition rate in complex environments. To address this issue, this paper proposes the Hierarchical-Concatenate Fusion(HCF)-TDNN model by adding HCF Module to ECAPA-TDNN model for sound event classification. In the HCF module, firstly, the audio signal is converted into two-dimensional time-frequency features for segmentation. Then, the segmented features are convolved one by one for improving the small receptive field in perceiving details. Finally, after the convolution is completed, the two adjacent parts are combined before proceeding with the next convolution for enlarging the receptive field in capturing large targets. Therefore, the improved model further enhances the scalability by emphasizing channel attention and efficient propagation and aggregation of feature information. The proposed model is trained and validated on the Urbansound8K dataset. The experimental results show that the proposed model can achieve the best classification accuracy of 95.83%, which is an approximate improvement of 5% (relatively) over the ECAPA-TDNN model.

ADD-YOLO: An algorithm for detecting animals in outdoor environments based on unmanned aerial imagery

The breeding management of extensive livestock and scientific research surveys of animals in outdoor environments often require the utilization of UAVs due to their ability to efficiently cover large areas at a cost-effective rate. However, identifying small animal targets in aerial imagery from high-altitudes remains a significant challenge. This paper introduces an enhanced algorithm based on YOLOv8n, specifically designed for aerial animal detection. Firstly, we add a P2 small target detection layer on top of the original baseline model, while removing the P5 large target detection layer and 32x downsampling to enhance the detection of small animal targets and reduce the number of model parameters. Secondly, The improved N-SPPCSPC module replaces the spatial pyramid pooling structure in the baseline model to enhance the extraction capability for small targets. Thirdly, an improved DWRC2f module is adopted to enhance the extraction of multi-scale contextual information. Fourthly, the SEAM module is incorporated before the detection head to enhance the detection of occluded and overlapping animals. Finally, a combined NWD Loss function is implemented to address the scale sensitivity of IoU Loss, thereby improving the accuracy of small target detection. Compared to the baseline model, the improved model achieved an increase of 7.1% and 4.9% in mAP50 values and an increase of 4.0% and 1.3% in mAP50-95 values, respectively, across two datasets, while significantly reducing the number of parameters. Further comparisons with other single-stage object detection models demonstrate a better robustness of our model. Additionally, after quantization, when testing the inference speed of ADD-YOLO on performance-constrained edge devices, it was 1.72 times and 1.81 times faster than the baseline model. Therefore, this model provides a new and efficient monitoring tool for extensive pastoral management and wildlife surveys.

GCS-YOLOv8: A Lightweight Face Extractor to Assist Deepfake Detection

To address the issues of target feature blurring and increased false detections caused by high compression rates in deepfake videos, as well as the high computational resource requirements of existing face extractors, we propose a lightweight face extractor to assist deepfake detection, GCS-YOLOv8. Firstly, we employ the HGStem module for initial downsampling to address the issue of false detections of small non-face objects in deepfake videos, thereby improving detection accuracy. Secondly, we introduce the C2f-GDConv module to mitigate the low-FLOPs pitfall while reducing the model’s parameters, thereby lightening the network. Additionally, we add a new P6 large target detection layer to expand the receptive field and capture multi-scale features, solving the problem of detecting large-scale faces in low-compression deepfake videos. We also design a cross-scale feature fusion module called CCFG (CNN-based Cross-Scale Feature Fusion with GDConv), which integrates features from different scales to enhance the model’s adaptability to scale variations while reducing network parameters, addressing the high computational resource requirements of traditional face extractors. Furthermore, we improve the detection head by utilizing group normalization and shared convolution, simplifying the process of face detection while maintaining detection performance. The training dataset was also refined by removing low-accuracy and low-resolution labels, which reduced the false detection rate. Experimental results demonstrate that, compared to YOLOv8, this face extractor achieves the AP of 0.942, 0.927, and 0.812 on the WiderFace dataset’s Easy, Medium, and Hard subsets, representing improvements of 1.1%, 1.3%, and 3.7% respectively. The model’s parameters and FLOPs are only 1.68 MB and 3.5 G, reflecting reductions of 44.2% and 56.8%, making it more effective and lightweight in extracting faces from deepfake videos.

Basolateral Mechanics Prevents Rigidity Transition in Epithelial Monolayers.

The mechanics of epithelial tissues, which is governed by forces generated in various cell regions, is often investigated using two-dimensional models that account for the apically positioned actomyosin structures but neglect basolateral mechanics. We employ a more detailed three-dimensional model to study how lateral surface tensions affect the structure and rigidity of such tissues. We find that cells are apicobasally asymmetric, with one side appearing more ordered than the other depending on target cell apical perimeter. In contrast to the 2D model, which predicts a rigidity transition at large target perimeters, tissues in the 3D model remain solidlike across all parameter space.

Mutation potentiates migration swamping in polygenic local adaptation.

Locally adapted traits can exhibit a wide range of genetic architectures, from pronounced divergence at a few loci to small frequency divergence at many loci. The type of architecture that evolves depends strongly on the migration rate, as weakly selected loci experience swamping and do not make lasting contributions to divergence. Simulations from previous studies showed that even when mutations are strongly selected and should resist migration swamping, the architecture of adaptation can collapse and become transient at high mutation rates. Here, we use an analytical two-population model to study how this transition in genetic architecture depends upon population size, strength of selection, and parameters describing the mutation process. To do this, we develop a mathematical theory based on the diffusion approximation to predict the threshold mutation rate above which the transition occurs. We find that this performs well across a wide range of parameter space, based on comparisons with individual-based simulations. The threshold mutation rate depends most strongly on the average effect size of mutations, weakly on the strength of selection, and marginally on the population size. Across a wide range of the parameter space, we observe that the transition to a transient architecture occurs when the trait-wide mutation rate is 10-3-10-2, suggesting that this phenomenon is potentially relevant to complex traits with a large mutational target. On the other hand, based on the apparent stability of genetic architecture in many classic examples of local adaptation, our theory suggests that per-trait mutation rates are often relatively low.

A Novel Method for Identifying Landslide Surface Deformation via the Integrated YOLOX and Mask R-CNN Model

The detection of landslide areas and surface characteristics is the prerequisite and basis of landslide hazard risk assessment. The traditional method relies mainly on manual field identification, and discrimination is based on the lack of unified quantitative standards. Thus, the use of neural networks for the quantitative identification and prediction of landslide surface deformation is explored. By constructing an integrated model based on YOLO X-CNN and Mask R-CNN, a deep learning-based feature detection method for landslide surface images is proposed. First, the method superimposes Unmanned Aerial Vehicle (UAV) oblique photography data (UOPD) and Internet heterosource image data (IHID) to construct a landslide surface image dataset and landslide surface deformation database. Second, an integrated model suitable for small- and medium-scale target detection and large-scale target edge extraction is constructed to automatically identify and extract landslide surface features and to achieve rapid detection of landslide surface features and accurate segmentation and deformation recognition of landslide areas. The results show that the detection accuracy for small rock targets is greater than 80% and that the speed is 57.04 FPS. The classification and mask segmentation accuracies of large slope targets are approximately 90%. A speed of 7.89 FPS can meet the needs of disaster emergency response; this provides a reference method for the accurate identification of landslide surface features.

Anisotropies related to representational gravity.

Four experiments examined whether representational gravity, in which memory for the location of a previously-viewed target is displaced in the direction of implied gravitational attraction, occurs uniformly across a target. Participants viewed stationary, vertically-moving, or horizontally-moving targets of different sizes and at different heights within the picture plane. After a target vanished, participants indicated the remembered location of the top edge or bottom edge of that target. Significant anisotropies were found, as the remembered location of the top edge was displaced downward, whereas the remembered location of the bottom edge was not displaced or was displaced upward. Anisotropies along the vertical axis were not influenced by whether participants knew prior to target presentation which edge to remember or by whether targets were stationary or moved vertically, although there was a trend foranisotropies along the vertical axis to be reduced when targets moved horizontally. Larger targets and targets higher in the picture plane resulted in larger displacement when targets were stationary, although effects of size and height were diminished when targets were moving. If the top edge and bottom edge of a target are considered analogous to the trailing edge and leading edge of a moving target, respectively, then anisotropies related to representational gravity are similar to anisotropies previously reported for representational momentum for horizontally-moving targets (as direction of implied gravitational attraction is downward). The existence of such anisotropies has implications for the representation of space and for the localization of and interaction with stimuli in the environment.

Dietary lipid is largely deposited in skin and rapidly affects insulating properties.

Skin has been shown to be a regulatory hub for energy expenditure and metabolism: mutations of skin lipid metabolism enzymes can change the rate of thermogenesis and susceptibility to diet-induced obesity. However, little is known about the physiological basis for this function. Here we show that the thermal properties of skin are highly reactive to diet: within three days, a high fat diet reduces heat transfer through skin. In contrast, a dietary manipulation that prevents obesity accelerates energy loss through skins. We found that skin was the largest target in a mouse body for dietary fat delivery, and that dietary triglyceride was assimilated both by epidermis and by dermal white adipose tissue. Skin from mice calorie-restricted for 3 weeks did not take up circulating lipids and showed a highly depleted stratum corneum. Dietary triglyceride acyl groups persist in skin for weeks after feeding. Using multi-modal lipid profiling, we have implicated both keratinocytes and sebocytes in the altered lipids which correlate with thermal function. In response to high fat feeding, wax diesters and ceramides accumulate, and triglycerides become more saturated. In contrast, in response to the dramatic loss of adipose tissue that accompanies restriction of the branched chain amino acid isoleucine, skin becomes more heat-permeable, resisting changes induced by Western diet feeding, with a signature of depleted signaling lipids. We propose that skin should be routinely included in physiological studies of lipid metabolism, given the size of the skin lipid reservoir and its adaptable functionality.

Comparing the impact of targeting limited driving pressure to low tidal volume ventilation on mortality in mechanically ventilated adults with COVID-19 ARDS: an exploratory target trial emulation

BackgroundAn association between driving pressure (∆P) and the outcomes of invasive mechanical ventilation (IMV) may exist. However, the effect of a sustained limitation of ∆P on mortality in patients with...

An Efficient Hybrid Method for Predicting RCS of Electrically Large Complex Targets

An Efficient Hybrid Method for Predicting RCS of Electrically Large Complex Targets

UWB RCS calculations of smooth targets via the multi-band beam summation method

We present an ultra-wide-band beam-summation (UWB-BS) algorithm for calculating the radar cross-section (RCS) of large smooth targets whose asymptotic numerical complexity is of O ( 1 ) . In this approach, the incident field is expanded into a sparse set of iso-diffracting Gaussian beams (ID-GB), tiling the phase space (PS) of beam origins and directions. The field of each ID-GB is then tracked through reflections from the scatterer until the beam emerges from the target zone and propagates to the far zone using a closed-form near-to-far (N2F) zone transformation, and finally, the beam contributions are aggregated to obtain the scattered field. The algorithm can accommodate a multi-octave frequency band by dividing it into octave-wide sub-bands and introducing a UWB-BS scheme such that the PS beam lattices in the higher sub-bands are obtained by spatial-spectral upsampling those of the lower bands. The PS summation of ID-GBs is a priori localized and involves a roughly constant relatively small number of contributing beams in all the sub-bands, leading to an O ( 1 ) (frequency-independent) asymptotic computational complexity. Furthermore, with the ID parametrization, the beam axes and their propagation and scattering parameters are frequency-independent and, therefore, need to be calculated only once and then reused for all frequencies, thus greatly reducing the numerical cost of computing the RCS at multiple frequencies. The paper describes all aspects of the algorithm, including the choice of the beam parameters for asymptotic error convergence and the utilization of localization for algorithmic efficacy.

Design and performance of 3D-Printed ABS@rGO/CF/CeO2 composites for microwave absorption and mechanical strength

Traditional microwave absorbing materials struggle to balance wideband absorption with mechanical properties, particularly in multifunctional applications. This study explores the effects of short carbon fibers (CF), cerium dioxide (CeO2), and reduced graphene oxide (rGO) on the electromagnetic and mechanical properties of 3D-printed composites. The ABS@CF/CeO2/rGO composite with a thickness of 2.2 mm exhibited exceptional microwave absorption, achieving an effective absorption bandwidth (EAB) of 6.6 GHz and a minimum reflection loss (RLmin) of −40.9 dB. The Whale Optimization Algorithm (WOA) was employed to design a load-bearing, microwave-absorbing superstructure, which significantly improved wideband absorption. Experimental results confirmed the superstructure’s isotropy and wide-angle absorption capabilities. The calculated electromagnetic parameters were applied to evaluate the radar cross-section (RCS) of large targets, demonstrating enhanced stealth performance for unmanned aerial vehicles (UAVs). Mechanical tests revealed that rGO improved interlayer bonding strength, while CF negatively impacted bending performance. The superstructure also showed excellent mechanical stability in load-bearing tests, underscoring its potential for practical engineering applications. This study offers insights into designing advanced composite materials with balanced electromagnetic and mechanical properties, guiding future multifunctional material development.

The production and separation of 161Tb with high specific activity at the University of Utah

Targeted radiotherapy (TRT) is an increasingly prominent area of research in nuclear medicine, particularly in the context of treating cancerous tumors. One radionuclide of considerable interest for TRT is terbium-161 (t1/2 = 6.95 days), which undergoes beta emission and shares similar decay properties as 177Lu (FDA-approved as LUTATHERA® and PLUVICTO®). Besides beta emission, 161Tb also emits a significant number of conversion and Auger electrons further enhancing its therapeutic potential. Terbium-161 can be produced using nuclear reactors through an indirect neutron capture reaction, G64160dn,γG64161d→3.66min,β−T65161b, from 160Gd targets. However, a key challenge in utilizing 161Tb for TRT lies in effectively separating target and product materials to attain high specific activity for radiolabeling. Here, we detail the production of no-carrier added 161Tb using low flux research reactors (mean thermal (<0.625 eV) neutron flux: 1.356×1012n∙cm−2∙s−1) like the University of Utah TRIGA Reactor, using enriched 160Gd2O3 targets (1.5 ± 0.3 μCi of 161Tb per mg of 160Gd target per hour of irradiation). We also developed a separation technique based on cation exchange and extraction chromatography, suitable for mCi level irradiations with targets exceeding 200 mg. In a simulated full-scale irradiation, 161Tb was successfully isolated from large mass targets using cation exchange (AG 50W-X8, with 2-hydroxyisobutyric acid at 70 mM, pH 4.75) and extraction chromatography (LN Resin, 0.5–0.75 M HNO3) methods. This resulted in high apparent molar activities of [161Tb]Tb-DOTA (113 ± 3 MBq/nmol), demonstrating high purity 161Tb relevant for potential future preclinical applications.

Investigate potential clinical benefits and linear energy transfer sparing utilizing proton arc therapy for hepatocellular carcinoma

PurposeTo investigate the potential clinical benefits and dose-averaged Linear Energy Transfer (LETd) sparing, utilizing proton arc plan for hepatocellular carcinoma (HCC) patients in comparison with Intensity Modulated Proton Therapy (IMPT). MethodsTen HCC patients have been retrospectively selected. Two planning groups were created: Proton Arc plans using Monaco ver. 6 and the clinical IMPT plan. Both planning groups used the same robustness parameters. The prescription dose is 67.5 Gy (RBE) in 15 fractions of the Clinical Target Volume (CTV). Robustness evaluations were performed to ensure dose coverage. Normal Tissue Complicated Probability (NTCP) model was utilized to predict the possibility of Radiation-Induced Liver Disease (RILD) and evaluate the potential benefit of proton arc therapy. LETd calculation and evaluation were performed as well. ResultsProton arc plan has shown better dosimetric improvements of most Organ-At-Risks (OARs). More specifically, the liver mean dose has been significantly reduced from 14.7 GyE to 10.62 GyE compared to the IMPT plan. The predicted possibility of RILD has also been significantly reduced for cases with a large and deep liver target where healthy liver tissue sparing is a challenge. Additionally, proton arc therapy could increase the average LETd in the target and reduce LETd in adjacent OARs. ConclusionsThe potential clinical benefit of utilizing proton arc therapy HCC varies depending on the patient-specific geometry. With more freedom, proton arc therapy can offer a better dosimetric plan quality in the challenge cases, which might not be feasible using the current IMPT technique.

Rapid Emergence of Resistance to Broad-Spectrum Direct Antimicrobial Activity of Avibactam.

Avibactam (AVI) is a diazabicyclooctane (DBO) β-lactamase inhibitor used clinically in combination with ceftazidime. At concentrations higher than those typically achieved in vivo , it also has broad-spectrum direct antibacterial activity against Enterobacterales strains, including metallo-β-lactamase-producing isolates, mediated by inhibition of penicillin-binding protein 2 (PBP2). This activity is mechanistically similar to that of more potent novel DBOs (zidebactam, nacubactam) in late clinical development. We found that resistance to AVI emerged readily, with a mutation frequency of 2×10 -6 to 8×10 -5 . Whole genome sequencing of resistant isolates revealed a heterogeneous mutational target that permitted bacterial survival and replication despite PBP2 inhibition, in line with prior studies of PBP2-targeting drugs. While such mutations are believed to act by upregulating the bacterial stringent response, we found a similarly high mutation frequency in bacteria deficient in components of the stringent response, although we observed a different set of mutations in these strains. Although avibactam-resistant strains had increased lag time, suggesting a fitness cost that might render them less problematic in clinical infections, there was no statistically significant difference in growth rates between susceptible and resistant strains. The finding of rapid emergence of resistance to avibactam as the result of a large mutational target has important implications for novel DBOs with potent direct antibacterial activity, which are being developed with the goal of expanding cell wall-active treatment options for multidrug-resistant gram-negative infections but may be vulnerable to treatment-emergent resistance.

Improving YOLOv7 for Large Target Classroom Behavior Recognition of Teachers in Smart Classroom Scenarios

Improving YOLOv7 for Large Target Classroom Behavior Recognition of Teachers in Smart Classroom Scenarios

Sputtering of targets in high-intensity heavy-ion beam experiments

Based on available models and experimental data, the sputtering of target atoms in long-term experiments with intense heavy ion (HI) beams has been considered. The experiments on the synthesis of superheavy nuclei (SHN), which are carried out in laboratories around the world, are examples of such experiments encountering the target atoms sputtering in specific conditions. Rotating wheels with a relatively large target annulus area are used in these experiments to reduce the temperature and radiation loads on the target. Large areas of rotating targets allow one to reduce the sputtering yields of target atoms significantly. The question arises about the reliability of these yields in experiments on the synthesis of SHN with Z>118. These nuclei can be produced with extremely low cross sections, requiring HI beam doses exceeding 10 20 particles passed through the target to observe several decay events of the thus synthesized SHN. Estimates based on approximations and simulations considered in this work are presented and compared with some data on actinide target sputtering obtained in experiments performed on the synthesis of SHN.

Lung SBRT: Dose gradient optimization based on target size

This study investigated optimization settings that steepen the dose gradient as a function of target size for lung stereotactic body radiation therapy (SBRT). Sixty-eight lung SBRT patients with planning target volumes (PTVs) ranging from 2-203 cc were categorized into small (<20 cc), medium (20-50 cc), and large (>50 cc) groups. VMAT plans were generated using the normal tissue objective (NTO) to penalize the dose gradient at progressively steeper NTO fall-off values (0.1, 0.2, 0.3, 0.4, 0.5 mm-1). Dose was calculated using the AcurosXB algorithm and was normalized so the prescription dose covered 95% of the PTV. Mann-Whitney, Kruskal-Wallis and ANOVA tests were used to assess for statistical differences in the Conformity Index at the 50% isodose level (CI50%), global maximum dose (Dmax), and monitor units (MU) across the various NTO settings. All plans adhered to institutional criteria and met the guidelines of the Radiation Therapy Oncology Group 0813. Steeper NTO fall-off values significantly increased Dmax and MUs across all groups (p < 0.05). CI50% significantly differed with fall-off values in small (0.3 mm-1) and medium (0.2 mm-1) targets, indicating steeper NTO fall-off values improve CI50% for small and medium targets (p < 0.05). Large targets showed no significant CI50% difference across these fall-off values. As target size increases, the importance of fall-off values in achieving an acceptable CI50% diminishes. Smaller targets benefit from steeper fall-off values despite increased Dmax and MUs. Consideration of fall-off value relative to target size is crucial to limit dose spillage outside the target.

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

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