Aiming at recognizing small proportion, blurred and complex traffic sign in natural scenes, a detection method based on RetinaNet-NeXt is proposed. First, to ensure the quality of dataset, data were cleaned enhanced denoise. Secondly, novel backbone network ResNeXt was employed improve accuracy effection RetinaNet. Finally, transfer learning group normalization adopted accelerate our training. Experimental results show that precision, recall mAP method, compared with original RetinaNet, are...

Semi-supervised learning has achieved significant success in the field of medical image segmentation. However, overfitting to erroneous pseudo-labels can lead cognitive biases models, a persistent issue semi-supervised that undermines its performance. To address issues above, we propose novel segmentation method named as Dual Task Correction Framework (DTCF). More specifically, Dual-Task Collaborative Review (DTCR) module, Spatial Perception Module (SPM), and Dynamic Pseudo-Label Generation...

Automatic segmentation of medical images is a crucial step for lesion measurement in computer-aided diagnosis. Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs) are widely adopted but have limitations. To address these challenges, we propose Frequency-enhanced Lightweight Mamba Network (FMamba) automatic image segmentation. Specifically, introduce the State Space (VSS) Frequency Feature Enhancement (FFE) modules efficient parallel feature extraction. The VSS module employs...

Accuracy segmentation of COVID-19 lesions in lung CT images can aid patient screening and diagnosis. However, the blurred, inconsistent shape location lesion area poses a great challenge to this vision task. To tackle issue, we propose multi-scale representation learning network (MRL-Net) that integrates CNN with Transformer via two bridge unit: Dual Multi-interaction Attention (DMA) Boundary (DBA). First, obtain local detailed feature global contextual information, combine low-level...

In this letter, the graph-based visual saliency (GBVS) model is extended by pulse-coupled neural network (PCNN) to implement well-defined criteria for a detector. receptive field, resized intensity feature map generated GBVS was regarded as input image of PCNN. After modulation, optimal iteration number and threshold were identified Otsu's method in pulse generator part, respectively. Moreover, other parameters PCNN set automatically. end, an automatic salient region detection algorithm...

Laplacian Biogeography-Based Optimization (LxBBO) is a BBO variant which improves BBO’s performance largely. When it solves some complex problems, however, has drawbacks such as poor performance, weak operability, and high complexity, so an improved LxBBO (ILxBBO) proposed. First, two-global-best guiding operator created for the worst habitat mainly to enhance exploitation of LxBBO. Second, dynamic two-differential perturbing proposed first two best habitats’ updating improve global search...

Lévy flight Shuffle Frog Leaping Algorithm (LSFLA) is a SFLA variant and enhances the performance of largely, however, it still has some defects, such as poor convergence low efficiency. So an improved LSFLA, namely, LSFLA based on Differential perturbation Quasi-Newton search (DQLSFLA), proposed in this paper. Firstly, way updating only one solution which worst at every sub-iteration replaced with all-solution subgroup to improve probability obtaining best solution, omit sorting step number...

Abstract Purpose Corona virus disease 2019 (COVID‐19) is threatening the health of global people and bringing great losses to our economy society. However, computed tomography (CT) image segmentation can make clinicians quickly identify COVID‐19‐infected regions. Accurate infection area COVID‐19 contribute screen confirmed cases. Methods We designed a network for regions in CT images. To begin with, multilayered features were extracted by backbone Res2Net. Subsequently, edge infected...

Single-modality medical images often cannot contain sufficient valid information to meet the requirements of clinical diagnosis. The diagnostic efficiency is always limited by observing multiple at same time. Image fusion a technique that combines functional modalities such as positron emission computed tomography (PET) and single-photon (SPECT) with anatomical (CT) magnetic resonance imaging (MRI) supplement complementary information. Meanwhile, fusing two (like CT-MRI) required replace...

The U-Net network has its own powerful capabilities in medical image segmentation tasks, yet still it is a challenging task to make accurately segment the infected lesions of COVID-19 CT images because these lesion areas are usually irregular shape, various size, and blurry boundaries. In this paper, novel multiscale U-shaped based on for accurate regions proposed. First, we generate two auxiliary scale features (fi0.5, fi1.5) main feature (fi1.0) through zoom strategy. Secondly, design...

The epidemic routing has been integrated into many applications, ranging from the worm propagation of online social networks to message diffusion in offline physical systems. Modelling provides a baseline evaluate system performance; it also becomes very desirable for engineers have theoretical guidance before they deploy real system. Early works analyse dynamics with average contact rate, i.e., each node will encounter same number other nodes time slot. They neglect status encountered...

The epidemic routing has been integrated into many applications, ranging from the worm propagation of online social networks to message diffusion in offline physical systems. Modelling provides a baseline evaluate system performance; it also becomes very desirable for engineers have theoretical guidance before they deploy real system. Early works analyse dynamics with average contact rate, i.e., each node will encounter same number other nodes time slot. They neglect status encountered...