Backbone Layers Research Articles

Lightweight prohibited items detection model in X-ray images based on improved YOLOv7-tiny

X-ray security inspection for prohibited item detection is widely used to maintain social order and personal safety. Deploying most modern object identification algorithms on portable security inspection devices is difficult because of their high computational complexity and many parameters. Consequently, an enhanced YOLOv7-tiny model is presented in this research and used to the problem of forbidden object recognition in X-ray images. First, the upgraded model’s feature extraction backbone is a lightweight network called FasterNet, which lowers computing complexity and speeds up detection. Second, the ELAN module in the original YOLOv7-tiny is replaced with a New-ELAN module with PConv in the neck layer to further reduce repeated computations and memory usage. Finally, to improve the detection performance of forbidden objects with smaller sizes, a lightweight CA attention mechanism is inserted between the neck and backbone layers. The studies are carried out using the SIXray dataset. The enhanced model decreases the volume, computational complexity, and number of parameters by 14.7%, 19.7%, and 12.2%, respectively, in comparison to the original YOLOv7-tiny model. The model achieves 88.7% identification accuracy for forbidden items, and the speed of detection rises from 82.4 FPS to 89.5 FPS. The experimental findings demonstrate that the improved model achieves overall lightweighting while maintaining good detection accuracy and speeding up detection. Code is available at https://github.com/zhg-SZPT/TinyRay.

Shutters-Inspired metal ions coordination hydrogel Strain/Pressure sensor for joint behavior evaluation and flatfeet correction

The metal ions coordination hydrogels (MI-HG) that polymer gels serve as flexible substrates and metal ions as conductive media have emerged as flexible electronic sensor materials. Nevertheless, fabricating MI-HG by a traditional soaking method is difficult to guarantee the good permeation of metal ions, manifesting as the “core–shell” structure of MI-HG, which undoubtedly reduces their mechanical property, conductivity, and sensitivity. Inspired by the shutter opening and closing regulation mechanism, a strategy that completely opening the backbone layers of polymer gels facilitates metal ions permeation thus improving the whole performance of MI-HG, especially the mechanical property is proposed. Herein, poly(vinyl alcohol)(poly(acrylic acid)-polyacrylamide) (PVA(PAA-PAM)) is used as the typical polymer hydrogel (HG), trivalent aluminum ion (Al3+) as the metal salt, and N,N'-methylenebisacrylamide (MBAA) as the important regulation ‘rod’, fabricating a series of aluminum coordination hydrogels (Al-HGx%). The results indicate that the mechanical performances of Al-HG2% are significantly improved. For example, the toughness of Al-HG2% reaches 996.12 KJ/m3, approximately three times that of the Al-HG0% sample (354.45 KJ/m3). The combination of structural characterization as well as finite element analysis, the mechanism for the enhanced mechanical performances based on shutter’s completely open effect is elucidated, the product of which (Al-HG2%) has potential application in flexible strain/pressure sensor for joint behavior evaluation and flatfeet correction.

SSMA-YOLO: A Lightweight YOLO Model with Enhanced Feature Extraction and Fusion Capabilities for Drone-Aerial Ship Image Detection

Due to the unique distance and angles involved in satellite remote sensing, ships appear with a small pixel area in images, leading to insufficient feature representation. This results in suboptimal performance in ship detection, including potential misses and false detections. Moreover, the complexity of backgrounds in remote sensing images of ships and the clustering of vessels also adversely affect the accuracy of ship detection. Therefore, this paper proposes an optimized model named SSMA-YOLO, based on YOLOv8n. First, this paper introduces a newly designed SSC2f structure that incorporates spatial and channel convolution (SCConv) and spatial group-wise enhancement (SGE) attention mechanisms. This design reduces spatial and channel redundancies within the neural network, enhancing detection accuracy while simultaneously reducing the model’s parameter count. Second, the newly designed MC2f structure employs the multidimensional collaborative attention (MCA) mechanism to efficiently model spatial and channel features, enhancing recognition efficiency in complex backgrounds. Additionally, the asymptotic feature pyramid network (AFPN) structure was designed for progressively fusing multi-level features from the backbone layers, overcoming challenges posed by multi-scale variations. Experiments of the ships dataset show that the proposed model achieved a 4.4% increase in mAP compared to the state-of-the-art single-stage target detection YOLOv8n model while also reducing the number of parameters by 23%.

Electrochemical investigation of Pr6O11 infiltration into La0.8Sr0.2MnO3-δ-Ce0.9Gd0.1O1.95 cathodes for IT-SOFC

Infiltration through spray-pyrolysis deposition is a one-step and effective method for enhancing the electrode performance of solid oxide fuel cells (SOFCs) at intermediate temperatures. Among various infiltrated electrodes, Pr6O11 has shown promising electrocatalytic properties for oxygen reduction reaction (ORR), despite its relatively low electrical conductivity. This study focuses on obtaining Pr6O11-infiltrated cathodes using spray-pyrolysis deposition, with different porous backbone layers that exhibit electronic and/or ionic conductivity: La0.8Sr0.2MnO3-δ (LSM), Ce0.9Gd0.1O1.95 (CGO), and LSM-CGO composite. The electrochemical properties and the subprocesses involved in the ORR are investigated in symmetrical cells. It is observed that the polarization resistance (Rp) decreases when the backbone layer exhibits high ionic conductivity. At 650 °C, the Rp values are 0.03, 0.06 and 0.1 Ω cm2 for CGO, LSM-CGO and LSM backbones, respectively. The dependence of Rp contributions with pO2 reveals that the same electrochemical processes are involved regardless of the backbone composition, confirming a less resistive oxide-ion transport from the TPB into the electrolyte with the CGO backbone. In contrast, the CGO backbone exhibits higher series resistance (Rs) (44.5 Ω cm), attributed to an increased contribution of ohmic losses compared to the LSM backbone (39.8 Ω cm). This study provides valuable insights into the performance and optimization of Pr6O11-infiltrated cathodes with different backbone compositions for their implementation in SOFCs.

KCPNet: Knowledge-Driven Context Perception Networks for Ship Detection in Infrared Imagery

Ship detection plays a crucial role in a variety of military and civilian marine inspection applications. Infrared images are irreplaceable data sources for ship detection due to their strong adaptability and excellent all-weather reconnaissance ability. However, previous researches mainly focus on visible light or synthetic aperture radar (SAR) ship detection, while infrared ship detection is left in a huge blind spot. The main obstacles to this dilemma lie in the absence of public datasets, small scale, and poor semantic information of infrared ships, and severe clutter in complex ocean environments. To address the above challenges, we propose a knowledge-driven context perception network (KCPNet) and construct a public dataset called infrared ship detection dataset (ISDD). In KCPNet, aiming at the small scale of infrared ships, a balanced feature fusion network (BFF-Net) is proposed to balance information from all backbone layers and generate nonlocal features with balanced receptive fields. Moreover, considering the key role of contextual information, a contextual attention network (CA-Net) is designed to improve robustness in complex scenes by enhancing target and contextual information and suppressing clutter. Inspired by prior knowledge of human cognitive processes, we construct a novel knowledge-driven prediction head to autonomously learn visual features and back-propagate the knowledge throughout the whole network, which can efficiently reduce false alarms. Extensive experiments demonstrate that the proposed KCPNet achieves state-of-the-art performance on ISDD. Source codes and ISDD are accessible at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/yaqihan-9898</uri> .

Multi-Dimensional Graph Interactional Network for Progressive Point Cloud Completion

Point cloud completion refers to inferring the complete and visually plausible shape from a partial input. Existing point cloud completion methods focus on recovering the global integrity of partial point clouds but lack local structural details. Furthermore, they seldom consider the shape faithfulness of completed results, that some completed points fail to fall into the ground-truth position faithfully. To meet the above challenges, we present a multidimensional graph interactional network for progressive point cloud completion. Specifically, we propose a multiresolution multidimensional graph encoder (MRMD-GE) to capture the information from both within-dimension and cross-dimension interactions for the purpose of enhancing the perception of local geometry. Inspired by the FPN, we develop a recursive point cloud pyramid decoder (RPPD) for generating multistage completed point clouds progressively, which incorporates extra feedback connections into the bottom-up backbone layers. In addition, we design a depth map discriminator combined with differentiable rendering to match the distribution of generated and real point clouds, making the completed point clouds more faithful to the ground truth. Quantitative and qualitative experiments on Completion3D, Shapenet-Part, and KITTI datasets demonstrate that our proposed method has compelling advantages over the state-of-the-art methods.

Direct utilization of gaseous fuels in metal supported solid oxide fuel cells

Direct utilization and internal reforming of gaseous fuels is investigated on symmetric-architecture metal supported solid oxide fuel cells (MS-SOFCs) with thin ceramic electrolyte and scaffold backbone layers, and low cost ferritic stainless steel supports on both sides. Infiltrated Pr-oxide and Ni/samarium-doped ceria catalysts are added to the cathode and anode electrodes, respectively. Initial performance and durability is evaluated for MS-SOFCs operating with natural gas, propane, ammonia, and dimethyl ether at 700 °C. Cells for natural gas and propane utilize a novel high entropy alloy (HEA) catalyst for internal reforming with performance and degradation rates similar to H2 (0.5 W cm−2 and ∼12%/100 h). Initial testing with sulfur shows reversible degradation for levels found in natural gas and irreversible degradation for higher levels found in commercial propane. Overall, MS-SOFCs show successful fuel flexibility.

Automated bridge crack detection method based on lightweight vision models

Deep learning-based bridge crack detection methods have advantages over traditional methods. We proposed an automated bridge crack detection method using lightweight vision models. First, our study applied the You Only Look Once 4th version (YOLO v4) (Bochkovskiy et al. in Yolov4: Optimal speed and accuracy of object detection. arXiv:200410934, 2020) to bridge surface crack detection. Then, to achieve model acceleration, some lightweight networks were used to replace the feature extraction network in YOLO v4, which reduced the parameter numbers and the backbone layers. The lightweight design can reduce the computational overhead of the model, making it convenient to deploy on edge platforms with limited computational power. The experimental results showed that the lightweight network-based bridge crack detection model required significantly less storage space at the expense of a slight reduction in precision. Therefore, an improved YOLO v4 crack detection method was proposed to meet real-time running without sacrificing accuracy. The precision, recall, and F1 score of the proposed crack detection method are 93.96%, 90.12%, and 92%, respectively. And the model only required 23.4 MB of storage space, and its frames per second could reach 140.2 frames. Compared with existing bridge crack detection methods, the proposed method showed precision, speed, and model size advantages.

Self-Balanced R-CNN for instance segmentation

Current state-of-the-art two-stage models on instance segmentation task suffer from several types of imbalances. In this paper, we address the Intersection over the Union (IoU) distribution imbalance of positive input Regions of Interest (RoIs) during the training of the second stage. Our Self-Balanced R-CNN (SBR-CNN), an evolved version of the Hybrid Task Cascade (HTC) model, brings brand new loop mechanisms of bounding box and mask refinements. With an improved Generic RoI Extraction (GRoIE), we also address the feature-level imbalance at the Feature Pyramid Network (FPN) level, originated by a non-uniform integration between low- and high-level features from the backbone layers. In addition, the redesign of the architecture heads toward a fully convolutional approach with FCC further reduces the number of parameters and obtains more clues to the connection between the task to solve and the layers used. Moreover, our SBR-CNN model shows the same or even better improvements if adopted in conjunction with other state-of-the-art models. In fact, with a lightweight ResNet-50 as backbone, evaluated on COCO minival 2017 dataset, our model reaches 45.3% and 41.5% AP for object detection and instance segmentation, with 12 epochs and without extra tricks. The code is available at https://github.com/IMPLabUniPr/mmdetection/tree/sbr_cnn.

Reliability Evaluation of Cyber-Physical Power Systems Considering Random Failures in Measurement and Remote Control

In integrated cyber-physical power systems (CPPS), the malfunctions of cyber system deeply affect the dispatcher’s remedial action scheme, and thus have a significant impact on the reliability of the physical power system. The random failures of electrical devices in physical power system and the random outages of measurement and remote control in cyber system are both considered for the CPPS reliability evaluation in this paper. Firstly, the component outage models in physical, access, backbone and application layers are presented. Then, a linearized DC state estimation (SE) model is introduced and the SE errors incurred by measurement failures, i.e. bad and lost measurements, are interpreted. Moreover, the impact of SE errors and remote control failures on the corrective action scheme of the control center during the fault recovery process is simulated by a novel modification of the classical optimal load shedding (OLS) model. Finally, to verify the effectiveness and validity of the proposed CPPS reliability evaluation method, the communication network topologies of the cyber system for the RBTS and IEEE-RTS96 are designed, and then the reliability evaluation based on the presented component outage models and the novel modified OLS model are conducted by the non-sequential Monte Carlo simulation (NMCS).

Detecting the Early Flowering Stage of Tea Chrysanthemum Using the F-YOLO Model

Detecting the flowering stage of tea chrysanthemum is a key mechanism of the selective chrysanthemum harvesting robot. However, under complex, unstructured scenarios, such as illumination variation, occlusion, and overlapping, detecting tea chrysanthemum at a specific flowering stage is a real challenge. This paper proposes a highly fused, lightweight detection model named the Fusion-YOLO (F-YOLO) model. First, cutout and mosaic input components are equipped, with which the fusion module can better understand the features of the chrysanthemum through slicing. In the backbone component, the Cross-Stage Partial DenseNet (CSPDenseNet) network is used as the main network, and feature fusion modules are added to maximize the gradient flow difference. Next, in the neck component, the Cross-Stage Partial ResNeXt (CSPResNeXt) network is taken as the main network to truncate the redundant gradient flow. Finally, in the head component, the multi-scale fusion network is adopted to aggregate the parameters of two different detection layers from different backbone layers. The results show that the F-YOLO model is superior to state-of-the-art technologies in terms of object detection, that this method can be deployed on a single mobile GPU, and that it will be one of key technologies to build a selective chrysanthemum harvesting robot system in the future.

OPMP: An Omnidirectional Pyramid Mask Proposal Network for Arbitrary-Shape Scene Text Detection

Scene text detection methods have achieved significant progresses. However, stack-omnidirectional text dilemma, under-segmentation of very close text words, and over-segmentation of arbitrary-shape long text lines, are still main challenges. Motivated by these problems, we proposed a two stage method called omnidirectional pyramid mask proposal text detector (OPMP). OPMP removes anchor mechanism that requires heuristic non-maximum suppress processing. Instead, it uses an effective pyramid lengthwise and sidewise residual sequence modeling method to produce arbitrary-shape proposals. To accurately extract the features of text shape, OPMP enhances the backbone layers by a multiple arbitrary-shape fitting mechanism. Finally, a multi-grain text classification module is proposed, which reclassifies each text region robustly. Comprehensive ablation studies demonstrate the effectiveness of each proposed component. In addition, experiments on various benchmarks, including ICDAR2015, MLT, MSRA-TD500, CTW1500, and Total-text, show that our method outperforms previous state-of-the-art methods.

Competition between Exceptionally Long‐Range Alkyl Sidechain Ordering and Backbone Ordering in Semiconducting Polymers and Its Impact on Electronic and Optoelectronic Properties

Intra- and intermolecular ordering greatly impact the electronic and optoelectronic properties of semiconducting polymers. Despite much prior efforts regarding molecular packing, the interrelationship between ordering of alkyl sidechains and conjugated backbones has not been fully detailed. We report here the discovery of a highly ordered alkyl sidechain phase in six representative semiconducting polymers, determined from distinct spectroscopic and diffraction signatures. The sidechain ordering exhibits unusually large coherence lengths of at least 70 nm, induces torsional/twisting backbone disorder, and results in a vertically layered multilayer nanostructure with ordered sidechain layers alternating with disordered backbone layers. Calorimetry and in-situ variable temperature scattering measurements in a model system PBnDT-FTAZ clearly delineate this competition of ordering that prevents the simultaneous long-range order of both moieties. The long-range sidechain ordering can be exploited as a transient state to fabricate PBnDT-FTAZ films with an atypical edge-on texture and 2.5x improved OFET mobility. The observed influence of ordering between the moieties implies that improved molecular design could produce synergistic rather than destructive ordering effects. Given the large sidechain coherence lengths observed, such synergistic ordering should greatly improve the coherence length of backbone ordering and thereby improve electronic and optoelectronic properties such as charge transport and exciton diffusion lengths.

Ba0.5Gd0.8La0.7Co2O6-δ Infiltrated in Porous BaZr0.7Ce0.2Y0.1O3 Backbones as Electrode Material for Proton Ceramic Electrolytes

Several porous backbone layers of proton conducting BaZr1-x-yCexYyO3-δ (BZCY) were prepared by spray-coating or painting on BaZr0.7Ce0.2Y0.1O2.95 (BZCY72) button cells followed by high temperature firing in air. This step enabled to produce ceramic backbones with various grain and pore sizes. The backbones were further infiltrated with mixed conducting double perovskite Ba0.5Gd0.8La0.7Co2O6-δ (BGLC587) suspensions and annealed in air. The resulting composite layers were tested as oxygen/steam side electrodes for proton ceramic fuel cells and electrolysers (PCFCs-PCEs). The results confirm that a high performing electrode material such as BGLC587 with partial proton conductivity and high thermal expansion can be applied on a proton conducting electrolyte with low thermal expansion by use of backbone infiltration without losing electrochemical functionality. Indeed, the electrodes display an apparent polarization resistance of only 0.03 Ω cm2 at 700°C in oxygen humidified with 2.7% H2O. We further extracted and parameterized the impedances associated with the charge and mass transfer reactions in a system where protons are the dominating charge carriers at intermediate to low temperatures and oxide ions dominate the overall transport at high temperature. The fitting revealed to what extent the charge and mass transfer reactions are short-circuited by electronic leak current across the sample at high temperatures and pO2’s. The acquired activation energies and pre-exponential values were used to explain materials-specific and micro-structural differences between the different electrode architectures.

Effect of side-chain asymmetry on the intermolecular structure and order-disorder transition in alkyl-substituted polyfluorenes.

We study relations among the side-chain asymmetry, structure, and order-disorder transition (ODT) in hairy-rod-type poly(9,9-dihexylfluorene) (PF6) with two identical side chains and atactic poly(9-octyl-9-methyl-fluorene) (PF1-8) with two different side chains per repeat. PF6 and PF1-8 organize into alternating side-chain and backbone layers that transform into an isotropic phase at T^{ODT}(PF6) and T_{bi}^{ODT}(PF1-8). We interpret polymers in terms of monodisperse and bidisperse brushes and predict scenarios T^{ODT}<T_{bi}^{ODT} and T^{ODT}∼T_{bi}^{ODT} for high and low grafting densities (the side-chain length above or below the average grafting distance). Calorimetry and x-ray scattering indicate the condition T^{ODT}(PF6)∼T_{bi}^{ODT}(PF1-8) following the low grafting prediction. PF6 side chains coming from the alternating backbone layers appear as two separate layers with thickness H(PF6), whereas PF1-8 side chains appear as an indistinguishable bilayer with a half thickness H_{bilayer}(PF1-8)/2≈H(PF6). The low grafting density region is structurally possible but not certain for PF6 and confirmed for PF1-8.

Differences in duplication age distributions between human GPCRs and their downstream genes from a network prospective

BackgroundHow gene duplication has influenced the evolution of gene networks is one of the core problems in evolution. Current duplication-divergence theories generally suggested that genes on the periphery of the networks were preferentially retained after gene duplication. However, previous studies were mostly based on gene networks in invertebrate species, and they had the inherent shortcoming of not being able to provide information on how the duplication-divergence process proceeded along the time axis during major speciation events.ResultsIn this study, we constructed a model system consisting of human G protein-coupled receptors (GPCRs) and their downstream genes in the GPCR pathways. These two groups of genes offered a natural partition of genes in the peripheral and the backbone layers of the network. Analysis of the age distributions of the duplication events in human GPCRs and "downstream genes" gene families indicated that they both experienced an explosive expansion at the time of early vertebrate emergence. However, we found only GPCR families saw a continued expansion after early vertebrates, mostly prominently in several small subfamilies of GPCRs involved in immune responses and sensory responses.ConclusionIn general, in the human GPCR model system, we found that the position of a gene in the gene networks has significant influences on the likelihood of fixation of its duplicates. However, for a super gene family, the influence was not uniform among subfamilies. For super families, such as GPCRs, whose gene basis of expression diversity was well established at early vertebrates, continued expansions were mostly prominent in particular small subfamilies mainly involved in lineage-specific functions.

FTIR spectroscopy of smectic elastomer films under lateral strain

Polarized Fourier transform infrared (FTIR) spectroscopy is used to study the strain‐induced compression of molecular layers in oriented smectic liquid crystal elastomer films. A reversible change of the smectic layer thickness in SmA and SmC* films in response to external strain was revealed earlier by optical reflectometry and X‐ray measurements. However, these methods cannot probe the mechanism of layer compression on a molecular level. Polarized FTIR spectra show that the induced mesogenic tilt, one of the possible mechanisms, is too small to provide the dominating contribution to the layer shrinkage. The FTIR absorbance spectra of stretched samples are also evidence that there are no significant changes of the order parameter. Apparently, layer compression is achieved by a certain interpenetration of neighbouring layers, and/or compression of the interstitial backbone and spacer layers.

Supramolecular Ordering, Thermal Behavior, and Photophysical, Electrochemical, and Electroluminescent Properties of Alkoxy-Substituted Yne-Containing Poly(phenylene−vinylene)s

Alkoxy-substituted and defect-free yne-containing poly(phenylene−vinylene)s, 15a−d, having the general constitutional unit (−Ph−C⋮C−Ph−CHCH−Ph−CHCH−)n synthesized through the olefination reactions of the fluorophoric dialdehyde 7 [1,2-bis(4-formyl-2,5-dialkoxyphenyl)acetylene] with various 2,5-dialkoxy-p-xylylenebis(diethylphosphonates), 14a−d, are reported. A new synthetic route to 1-bromo-2,5-dialkoxybenzaldehyde (starting material for the synthesis of 7) allowing the grafting of all types of alkoxy side chains is also reported. Bulk state properties of the four polymeric materials have been analyzed using various experimental methods giving information on the thermal behavior as well as self-assembling morphologies at different size scales. Crystalline superstructures of either spherulitic or rodlike morphology, comprised of polymeric backbone layers separated by the side chains are identified to develop by nucleation and growth process for symmetrically alkoxy substituted 15a−c. Discrepancy between op...

Direct electrochemistry of hemoglobin in biomembrane-like DHP–PDDA polyion-surfactant composite films

Hb–DHP–PDDA films were assembled by incorporating hemoglobin (Hb) into polyion-surfactant DHP–PDDA composite films on pyrolytic graphite (PG) electrodes, where DHP is an anionic surfactant dihexa decylphosphate, and PDDA is a polycation poly(diallyldimethylammonium). Cyclic voltammetry of Hb–DHP–PDDA films showed a pair of stable and reversible peaks for Hb Fe(III)/Fe(II) redox couple at about −0.34 V versus saturated calomel electrode (SCE) in pH 7.0 buffers. The electron transfer rate between Hb and PG electrodes was greatly facilitated in microenvironment of the films. X-ray diffraction (XRD) and differential scanning calorimetry (DSC) suggest that in both DHP–PDDA and Hb–DHP–PDDA films, DHP is self-assembled into an ordered bilayer structure which is sandwiched by PDDA backbone layers. Positions of Soret absorption band indicate that Hb keeps its secondary structure similar to its native state in the films in the medium pH range. Hb could act as an enzymatic catalyst in DHP–PDDA films to catalyze reduction of oxygen, trichloroacetic acid, and nitrite with significant decrease of overpotential.

Atomic force microscopic studies of thin oriented films of ferroelectric liquid crystalline elastomers with varying network architecture

Cross-linked ferroelectric liquid crystalline (FLC) polymers were studied by atomic force microscopy (AFM). Polysiloxane copolymers were synthesized with mesogenic and photo-cross-linkable side groups, the latter connected either directly to the backbone via a short spacer or as a terminal group on a part of the mesogens. These elastomers were prepared as thin, freely suspended films in homeotropic orientation. Topographic measurements depict plateaus separated by steps of characteristic height corresponding to the surfaces and edges of smectic layers. From the temperature behavior, as well as from the reaction to mechanical deformation, a model for the network architectures is proposed. In the first case (“intralayer cross-linking”), a predominantly two-dimensional network is formed within the backbone layers separating the smectic layers; in the second case (“interlayer cross-linking”), a three-dimensional network is established that is dependent on the mesophase present during cross-linking.