Building-change detection underpins many important applications, especially in the military and crisis-management domains. Recent methods used for change have shifted towards deep learning, which depends on quality of its training data. The assembly large-scale annotated satellite imagery datasets is therefore essential global building-change surveillance. Existing almost exclusively offer near-nadir viewing angles. This limits range changes that can be detected. By offering larger...

Semantic segmentation of remote sensing images (SSRSI), which aims to assign a category each pixel in images, plays vital role broad range applications, such as environmental monitoring, urban planning, and land resource utilization. Recently, with the successful application deep learning sensing, substantial amount work has been aimed at developing SSRSI methods using models. In this survey, we provide comprehensive review SSRSI. Firstly, current mainstream semantic models based on...

Abstract Background: Although the outcome of patients with gastric cancer (GC) has improved significantly recent implementation annual screening programs. Reliable prognostic biomarkers are still needed due to disease heterogeneity. Increasing pieces evidence revealed an association between immune signature and GC prognosis. Thus, we aim build immune-related that can estimate prognosis for GC. Methods: For identification a gene (IRGS), expression profiles clinical information were collected...

Object detection in Remote Sensing (RS) has achieved tremendous advances recent years, but it remains challenging for rotated object due to cluttered backgrounds, dense arrangements and the wide range of size variations among objects. To tackle this problem, Dense Context Feature Pyramid Network (DCFPN) a power α-Gaussian loss are designed paper. The proposed DCFPN can extract multi-scale information densely accurately by leveraging multi-path dilation layer cover all sizes objects remote...

As intelligent technologies and unmanned systems develop rapidly, the concept of cross-domain cooperative technology swarm has emerged, received widespread attention, gradually become high ground in competition system among countries worldwide.Based on development demand for China, this study summarizes research status crossdomain typical scenarios such as sea -air, air -ground, -ground/sea -ground thoroughly analyzes current status, technological demand, key directions...

Object recognition, which is also referred as object classification or type aims at discriminating types in remote sensing images. With the availability of high resolution images, recognition attracts more and attention. Different from traditional methods mainly using hand-crafted features, we propose an method that combines deep features extracted a convolutional neural network (CNN) to recognize aircrafts ships The proposed consists two stages. In training stage, images objects with...

In this study, we propose an on-site electrochemical platform for sensitive simultaneous genotyping of the two major EGFR mutations (19del and L858R) through plasma ctDNA based on tetrahedral DNA nanostructure decorated screen-printed electrodes.

Rotating object detection in remote sensing image has made substantial progress. However, the widely used feature extraction and fusion modules cannot accurately represent spatial location angle information of objects arbitrary orientation. A scheme that expresses characteristics is necessary when are densely distributed. This article proposes an effective network for rotating optical (BDR-Net). First, a aggregation module (FAM) designed to refine objects, which can generate refined maps....

Airplane detection in remote sensing images is a challenging task due to the diversity of airplanes and complexity backgrounds. In this paper, we propose an airplane method using convolutional neural networks (CNNs) coarse-to-fine manner which simulates image analysts. Our proposes coarse candidate regions containing multiple first, then finely detects each these regions. According manner, design precise efficient framework consists two CNNs with same structure. One CNN used coarsely...

Inshore ship detection in remote sensing images is a challenging task because of the connectivity and similarity between ships backgrounds. The usual shape feature not always applicable sometimes it hard to be extracted. In this paper, deep features extracted from convolutional neural network (CNN) are used for inshore detection. order feed CNN with exclusively positive negative samples, novel parallelogram image cropping (PIC) method proposed. traditional can only generate rectangle samples...

Existing aircraft recognition methods usually regard as an isolated classification problem, supposing that detection has finished. These use image slices each containing single input, which is often not the case in practice. In order to recognize remote sensing images contain multiple objects and background, we propose a human-computer fusion framework combines advantages of human computer. First, candidate using eye tracking, making efficient accurate search ability human. Then, two-step...

Most of the current object detection algorithms use pretrained models that are trained on ImageNet and then fine-tuned in network, which can achieve good performance terms general detectors. However, field remote sensing image detection, as significantly different from data, it is meaningful to explore a train-from-scratch technique for images. This paper proposes an framework scratch, SRS-Net, describes design densely connected backbone network provide integrated hidden layer supervision...

Image inpainting refers to synthesizing plausible contents for images with missing regions. However, current methods often create blurry textures, distorted structures and loss of details, especially when the image has complex scenes or large We propose a fine-grained adversarial model super resolution. It performs coarse-to-fine procedure in two stages. The proposed generator first synthesizes initial predictions regions novel encoder-decoder structure. Then it refines predicted by...

Scene classification is a key issue in the field of remote sensing image analysis. In recent years, lot scene methods have emerged due to convenient acquisition high spatial resolution images. existing literature, feature-level fusion method widely used improve performance. this paper, we propose decision-level based on convolutional neural networks (CNNs) for classification. The proposed accomplishes task through two steps. first step, CNN fine-tuned using training samples, and soft-max...

Remote sensing images often suffer from shadow duo to partially or totally occludes direct light an illumination source. In this paper, we propose a novel removal algorithm. The trimap could be generated automatically by morphological subtraction method according the result of detection. Then, weighted color and texture sample selection image matting is applied in order obtain accurate coefficient. experiment results are illustrated with practical examples verify efficacy

Introduction: Regulatory T cells (Tregs) play important roles in tumor immunosuppression and immune escape. The aim of the present study was to construct a novel Tregs-associated biomarker for prediction tumour microenvironment (TIME), clinical outcomes, individualised treatment hepatocellular carcinoma (HCC). Methods: Single-cell sequencing data were obtained from three independent cohorts. Cox LASSO regression utilised develop Tregs Related Scoring System (TRSSys). GSE140520, ICGC-LIRI...

With the convenient acquisition of high spatial resolution remote sensing images, scene classification has attracted great attention in recent years, leading to boom methods. Among these methods, feature-level fusion strategies are widely used. Different from existing we propose a decision-level method for using features extracted convolutional neural networks (CNNs). The task is accomplished at decision level through two steps. In first step, top-N possible classes test sample obtained...

This paper presents a new method for object detection by edge grouping. can detect the boundaries of objects under complex background where contours are partly occluded or missing during contour extraction. Our is adapted to with not only closed but also open-boundaries. There three contributions in this work. First, shape an represented novel Turn Angle Probabilistic Sequence Model (TAPSM) which originates from HMM. model robust noisy images. Second, grouping defined sequential search...

Object detection in remote sensing has developed rapidly and been applied many fields, but it is known to be vulnerable adversarial attacks. Improving the robustness of models become a key issue for reliable application deployment. This paper proposes robust object detector images (RSIs) mitigate performance degradation caused by For objects, multi-dimensional convolution utilized extract both specific features consistency from clean dynamically efficiently. enhances feature extraction...

Cancer-associated fibroblasts (CAFs) regulate the malignant biological behaviour of hepatocellular carcinoma (HCC) as a significant component tumour immune microenvironment (TIME). This study aimed to develop CAFs-based scoring system predict prognosis and TIME patients with HCC.Data for TCGA-LIHC GSE14520 cohorts were downloaded from The Cancer Genome Atlas Gene Expression Omnibus databases. Single-cell RNA-sequencing data HCC samples retrieved GSE166635 cohort. Least Absolute Shrinkage...

A novel method based on de-chirping technique for compressing stepped-frequency chirp signal (SFCS) was firstly introduced, and then applied to imaging simulation a spacecraft by ground ISAR using SFCS. The echo model an orbiting (the Chinese Shenzhou Spacecraft) analyzed deriving the equivalent translational rotational movement. Result showed that clearly imaged with resolution as high about 0.20 m