Weakly supervised image semantic segmentation has become the most popular method in recent years because of its low cost and has been widely used in medical image segmentation, automatic driving, remotesensingimage ...
详细信息
ISBN:
(纸本)9789819784929;9789819784936
Weakly supervised image semantic segmentation has become the most popular method in recent years because of its low cost and has been widely used in medical image segmentation, automatic driving, remotesensingimage analysis and other fields. However, the current weakly supervised semantic segmentation based on transformer has some problems, such as focusing on the whole, ignoring local details and confusing different categories. To solve these problems, we come up with a token-guided single stage weakly supervised image semantic segmentation algorithm. First of all, in order to solve the problem of insufficient attention to details, we proposed an optimization clipping method, which realized the selection of uncertain regions as much as possible and the fine marking of uncertain regions. Then, the single-class token to multiple class tokens method is purposed to obtain multiple class tokens for fine guidance. In particular, we designed a multiple class tokens guide method to complete the function of classifying uncertain regions and correctly activating them. The quantitative and qualitative results of the public dataset PASCAL VOC 2012 validate the effectiveness of the method.
The proceedings contain 19 papers. The topics discussed include: enhancing cyberattack detection in IoT environments through advanced resampling techniques;segmentation of hepatocytes nuclei using YOLO and mathematica...
ISBN:
(纸本)9798350391886
The proceedings contain 19 papers. The topics discussed include: enhancing cyberattack detection in IoT environments through advanced resampling techniques;segmentation of hepatocytes nuclei using YOLO and mathematical morphology;music genre classification using contrastive dissimilarity;a generative approach to electrical impedance tomography image reconstruction using prior information;on the possibilities of using traditional patternrecognition techniques in Brazilian banknote characterization;neural architecture search for enhancing action video recognition in compressed domains;denoising autoencoder for partial discharge identification in instrument transformers;and immersive virtual conference room: bridging the gap between remote collaboration and physical presence.
Spectral images include rich spatio-spectral information of target scene, which can accurately identify and distinguish the features of ground objects. Therefore, spectral image classification is widely used in remote...
详细信息
In view of the application of optical remotesensing disaster emergency rescue under the complex environment of low illumination at night, the optical remotesensing imaging technology under the condition of low illum...
详细信息
ISBN:
(数字)9781510667044
ISBN:
(纸本)9781510667037
In view of the application of optical remotesensing disaster emergency rescue under the complex environment of low illumination at night, the optical remotesensing imaging technology under the condition of low illumination and low signal-to-noise ratio was studied. The compressed sensing technology of thin film diffraction grating primary mirror is used to realize large-aperture optical acquisition. The sensing ability of large dynamic range is increased by Geiger pattern imaging technology, and the dim and weak targets are identified by semantic sensing algorithm. The system realizes target recognition under the condition of extremely low image signal-to-noise ratio through the design of the new system's main mirror flattening and the aliasing compression and decoupling of spatial information and spectral information. The technology has completed space-based scheme design and desktop principle verification tests, and the spectral resolution reaches 5nm, realizing fast target search and recognition.
Recently, object segmentation of remotesensingimages has achieved great progress in many fields, such as transportation, natural resource, ecology, et al. A lot of works mainly performed object segmentation in fully...
详细信息
Hyperspectral image (HSI) classification aims at assigning a unique label for every pixel to identify categories of different land covers. Existing deep learning models for HSIs are usually performed in a traditional ...
详细信息
ISBN:
(纸本)9798350301298
Hyperspectral image (HSI) classification aims at assigning a unique label for every pixel to identify categories of different land covers. Existing deep learning models for HSIs are usually performed in a traditional learning paradigm. Being emerging machines, quantum computers are limited in the noisy intermediate-scale quantum (NISQ) era. The quantum theory offers a new paradigm for designing deep learning models. Motivated by the quantum circuit (QC) model, we propose a quantum-inspired spectral-spatial network (QSSN) for HSI feature extraction. The proposed QSSN consists of a phase-prediction module (PPM) and a measurement-like fusion module (MFM) inspired from quantum theory to dynamically fuse spectral and spatial information. Specifically, QSSN uses a quantum representation to represent an HSI cuboid and extracts joint spectral-spatial features using MFM. An HSI cuboid and its phases predicted by PPM are used in the quantum representation. Using QSSN as the building block, we further propose an end-to-end quantum-inspired spectral-spatial pyramid network (QSSPN) for HSI feature extraction and classification. In this pyramid framework, QSSPN progressively learns feature representations by cascading QSSN blocks and performs classification with a softmax classifier. It is the first attempt to introduce quantum theory in HSI processing model design. Substantial experiments are conducted on three HSI datasets to verify the superiority of the proposed QSSPN framework over the state-of-the-art methods.
In order to improve the detection performance of traditional single-source remotesensingimage ship detection methods in terms of anti-interference capability and multi-scale targets. We develop an improved algorithm...
详细信息
Morphological processing has found several applications in image analysis and patternrecognition. Some of these techniques, known as morphological reconstruction algorithms, have been employed for land cover classifi...
详细信息
Morphological processing has found several applications in image analysis and patternrecognition. Some of these techniques, known as morphological reconstruction algorithms, have been employed for land cover classification in remotesensing data. In this paper, we analyse the mathematical foundations, applications, and limitations of reconstruction by dilation and by erosion oriented to urban extraction, using Sentinel-2 satellite data. Different techniques oriented to the proper determination of the marker and mask images, the basis for reconstruction, are proposed in this manuscript. In addition, in order to diminish the long computation time required for reconstruction, two parallel implementations using Multi-core and GPU, are proposed. According to our research, these algorithms can be considered as effective and non-supervised solutions for urban extraction applications based on multispectral satellite imagery.
remotesensing change detection (CD) is to use multitemporal remotesensing data to extract change information by using a variety of imageprocessing and patternrecognition methods, and quantitatively analyze and det...
详细信息
remotesensing change detection (CD) is to use multitemporal remotesensing data to extract change information by using a variety of imageprocessing and patternrecognition methods, and quantitatively analyze and determine the characteristics and processes of surface changes. In recent research on CD, how to more accurately segment objects and how to extract and effectively link spatiotemporal information are important parts. To achieve this, we propose a progressive sampling (PS) transformer network for remotesensingimage CD, which continuously extracts and optimizes feature information in an iterative manner, so that pixels can establish better connections in the spatial domain to model the context. Our intuition is that, through this iterative sampling method, the parts of interest in the image can be gradually extracted. This allows subsequent processing to be more focused on useful areas, which in turn reduces interference from uninteresting parts, and the information after PS will be generalize into several tokens containing rich semantic information. Using the excellent modeling ability of the transformer, the optimized tokens are mapped back to the original image features to achieve the purpose of segmenting accurate difference images. We conducted extensive experiments on three CD datasets, LEVIR-CD, DSIFN-CD, and WHU-CD, and achieved evaluation scores of 90.73/84.11, 80.10/68.93, and 91.67/85.15 on F1-score and IoU metrics, respectively. Notably, the convolutional neural network (CNN) backbone of our network uses only a simplified ResNet model, without using structurally complex frameworks, such as FPN and Unet, but our model uses PS module and transformer to achieve better performance than the recent advanced CD models.
To effectively remove noise from the point clouds obtained for road surface defect detection while preserving the feature information, a bilateral filtering algorithm based on curvature features is proposed for denois...
详细信息
暂无评论