Building NAS: Automatic designation of efficient neural architectures for building extraction in high-resolution aerial images

Jing, Weipeng; Lin, Jingbo

doi:10.1016/j.imavis.2020.104025

Cited by 8 publications

(5 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…FasterSeg [53] combines NAS with knowledge distillation to search a teacher and a student networks simultaneously. In [54], a jagged path pruning strategy was proposed to ignore invalid paths to reduce the search cost.…”

Section: Related Workmentioning

confidence: 99%

Efficient pyramid context encoding and feature embedding for semantic segmentation

Liu

Yin

2021

Image and Vision Computing

View full text Add to dashboard Cite

Section: Related Workmentioning

confidence: 99%

Efficient pyramid context encoding and feature embedding for semantic segmentation

Liu

Yin

2021

Image and Vision Computing

View full text Add to dashboard Cite

“…Super-resolution is one of the research hotspots in the field of computer vision, and it is widely used in remote sensing of satellite images, 3 – 5 monitoring equipment, 6 and other fields. In addition to improving the perceptual quality of images, super-resolution can also help improve other computer vision tasks such as image segmentation, 7 , 8 pose estimation, 9 and so on.…”

Section: Introductionmentioning

confidence: 99%

“…Super-resolution is one of the research hotspots in the field of computer vision, and it is widely used in remote sensing of satellite images, [3][4][5] monitoring equipment, 6 and other fields.…”

Section: Introductionmentioning

confidence: 99%

Synthetic aperture imaging using multi-view super-resolution

et al. 2023

View full text Add to dashboard Cite

.The occlusion problem is a major challenge in the field of computer vision. Synthetic aperture imaging (SAI) is often used for surface reconstruction of occluded objects. However, SAI usually relies on high-speed information transmission devices. In addition, a large amount of information in the scene is lost, when handling low-resolution input images. This limitation results in unclear reconstructed regions in the synthetic aperture image and thus hinders the application of SAI in downstream tasks. We propose a multi-view super-resolution SAI method. It aims to generate high-resolution synthetic aperture images using images acquired by few of low-resolution acquisition devices. The main contributions of this paper are: (1) a multi-view super-resolution algorithm is proposed. It can generate clear synthetic aperture images in an array with a limited number of cameras. (2) By exploiting the correlation between views, the proposed algorithm can generate super-resolution synthetic aperture images with more accurate image structure and sharper image edges. (3) A feature extraction module is proposed. It can effectively extract the complementary relationship between pictures from different perspectives. The experimental results show that the proposed method can generate a reconstructed image of the occluded object surface with clear edges and accurate structure. Compared to conventional SAI, our method improves 5.7 % / 21.1 % on peak signal-to-noise ratio (PSNR)/structure similarity index measure (SSIM) and 4.4 % / 9.2 % on PSNR/SSIM respectively on two datasets compared to other state-of-the-art super-resolution methods.

show abstract

“…As a solution, residual learning has been introduced in Ref. [30] for image recognition, and in some other works for generating deep neural networks [31]. For example, in Ref.…”

Section: Introductionmentioning

confidence: 99%

Spatial and polarimetric information fusion using residual network for polarimetric synthetic aperture radar image classification

Imani

2022

IET Radar Sonar & Navi

View full text Add to dashboard Cite

A polarimetric synthetic aperture radar (POLSAR) system provides an image that can be considered as a data cube containing spatial information in two spatial dimensions and polarimetric information in the scattering dimension. A spatial and polarimetric residual network (SPRN) is proposed for POLSAR image classification. At first, polarimetric features are extracted from the scattering dimension through two designed polarimetric residual blocks. Then, the processed POLSAR cube is fed to two consecutive spatial residual blocks for contextual feature extraction. Three dimensional convolutional layers are used as basic layers for simultaneous extraction and fusion of polarimetric information and correlation among neighbouring pixels in local regions. The shortcut connections are utilised to overcome the degradation problem due to increasing network depth. In addition, batch normalisation is applied to regularise the learning process. The experimental results on four real POLSAR images show the superior performance of SPRN compared to several state‐of‐the‐art classifiers in terms of various assessment measures.

show abstract

Building NAS: Automatic designation of efficient neural architectures for building extraction in high-resolution aerial images

Cited by 8 publications

References 9 publications

Efficient pyramid context encoding and feature embedding for semantic segmentation

Efficient pyramid context encoding and feature embedding for semantic segmentation

Synthetic aperture imaging using multi-view super-resolution

Spatial and polarimetric information fusion using residual network for polarimetric synthetic aperture radar image classification

Contact Info

Product

Resources

About