Synthetic aperture radar (SAR) plays a crucial role in maritime surveillance due to its capability for all-weather, all-day operation. However, SAR ship recognition faces challenges, primarily the imbalance and inadequacy of samples publicly available datasets, along with presence numerous outliers. To address these issues, this paper proposes classification method based on text-generated images tackle dataset imbalance. Firstly, an image generation module is introduced augment data. This...

Despite that deep learning has achieved state-of-the-art performance for medical image segmentation, its success relies on a large set of manually annotated images training are expensive to acquire. In this paper, we propose an annotation-efficient framework segmentation tasks avoids annotations images, where use improved Cycle-Consistent Generative Adversarial Network (GAN) learn from unpaired and auxiliary masks obtained either shape model or public datasets. We first the GAN generate...

Change detection based on synthetic aperture radar (SAR) images is an important application in the remote-sensing technology field. However, lack of labeled data has been a difficult problem SAR image detection, especially for pixel-level change detection. In this letter, we propose novel unsupervised algorithm, which improves accuracy by exploring features from both spatial and frequency domains images. particular, first clustering used as preclassification to obtain pseudo-labels then...

A new method in which the consensus algorithm is used to solve coordinate control problems of leaderless multiple autonomous underwater vehicles (multi-AUVs) with double independent Markovian switching communication topologies and time-varying delays among sensors investigated. This accomplished by first dividing topology into two different parts, i.e., velocity position, reduce data capacity per package sent between multi-AUVs ocean. Then, state feedback linearization simplify rewrite...

Bacterial extracellular vesicles (bEVs) secreted by Gram-negative bacteria are referred to as outer membrane (OMVs) because they originate in the membrane. OMVs membrane-coated 20–250 nm size. They contain lipopolysaccharide (LPS), peptidoglycan, proteins, lipids, nucleic acids, and other substances derived from their parent participate transmission of information host cells. have broad prospects terms potential application fields adjuvants, vaccines, drug delivery vehicles. Currently, there...

Multiple-object tracking is an important task in automated video surveillance. In this paper, we present a multiple-human-tracking approach that takes the single-frame human detection results as input and associates them to form trajectories while improving original by making use of reliable temporal information closed-loop manner. It works first forming tracklets, from which extracted, then refining responses inside also improves accuracy tracklets' quantities. After this, local...

Object tracking based on RGB images may fail when the color of tracked object is similar to that background. Hyperspectral with rich spectral features can provide more information for RGB-based trackers. However, there no fusion algorithm hyperspectral and images. In this paper, we propose a reliability-guided aggregation network (RANet) tracking, which guides combination through modality reliability improve performance. Specifically, dual branch Transformer Tracking (TransT) structure...

Analysis of the morphology and function right ventricle (RV) can be used for prediction diagnosis cardiovascular disease. Accurate description structure heart provided by analyzing cardiac magnetic resonance imaging (MRI) images. Noise interference intensity inhomogeneity MRI images addressed using a local clustering (LIC) model. However, segmentation RV in still remains challenge mainly due to its ill-defined borders. To address such challenge, an algorithm segmenting based on motion (LMIC)...

Despite wearable electronic devices having enormous potential applications, the recording quality has not sufficiently met requirements for fetal disease detection. In this article, we aim to propose a novel framework electrocardiography (ECG) extraction, where hybrid approach mainly considers combination of adaptive maternal QRS removal (AQRSR) algorithm and tensor-based joint blind source separation (JBSS) approach. To adapt dynamic change ECG signal suppress powerful heartbeat, AQRSR is...

In this paper, the problem of human detection in crowded scenes is formulated as a maximum posteriori problem, which, given set candidates, predefined 3-D shape models are matched with image evidence, provided by foreground extraction and probability boundary, to estimate configuration. The optimal solution obtained decomposing mutually related candidates into unoccluded occluded ones each iteration according graph description candidate relations then only matching for candidates. A...

Model repair is regarded as a new process mining technique. It can models of information systems according to the generated event logs. The existing approaches focus on eliminating deviations between activities performed by model and those observed in However, they cannot efficiently enhance other conformance dimensions, such precision simplicity. In order obtain repaired with higher conformance, this work designs approach. builds via Petri nets (PNs) investigates how PN-based choice...

Tracking live fish in an open underwater environment to investigate their behavior is of great value for many applications, e.g. biological and robotic research. However, tracking real world a challenging task due complex non-rigid deformation abrupt movement fish. In this paper, we explore incorporate motion property propose real-time method based on novel adaptive multi-appearance models strategy, which can be adapted various changes the appearance caused by deformation. Experimental...

Perceptual video coding (PVC) can provide a lower bitrate with the same visual quality compared traditional H.265/high efficiency (HEVC). In this work, novel H.265/HEVC-compliant PVC framework is proposed based on saliency model. Firstly, both an effective and efficient spatiotemporal model used to generate map. Secondly, perceptual scheme developed A saliency-based quantization control algorithm reduce bitrate. Finally, simulation results demonstrate that shows its superiority in objective...

Nonlinear blind source separation (BSS) is one of the unsolved problems in unsupervised learning, because solutions are highly non-unique when there no prior information for mixing functions. In this paper, we present a novel approach to tackle ill-posedness nonlinear BSS problem with few assumptions. The derivation our algorithm inspired by idea an efficient layer-by-layer representation approximate nonlinearity. Once such built, final output layer constructed solving convex optimization...

Pedestrian detection is a challenging task for video surveillance. The problem becomes more difficult when occlusion prevalent. In this paper, we extend deformable part-based pedestrian detector to in crowded scenes by considering both body part responses and detections' mutual spatial relationship. Specifically, first decompose the full into several detectors, whose can be computed efficiently from response of detector. Then, given hypotheses are nominated scores responses' Finally, local...

Chronic postoperative pain poses challenges, emphasizing the importance of accurately predicting in advance. Generally, perception is associated with temporal dynamics brain, which can be represented by microstates. Specifically, microstates are transient and patterned brain topographies formed temporally overlapping spatially synchronized oscillatory activities. Consequently, characterizing activity, offer valuable insights into perception.In this prospective study, 66 female patients...