Feature-Selection High-Resolution Network With Hypersphere Embedding for Semantic Segmentation of VHR Remote Sensing Images

Xu, Hanwen; Tang, Xinming; Ai, Bo; Yang, Fanlin; Wen, Zhen; Yang, Xiaomeng

doi:10.1109/tgrs.2022.3183144

Cited by 10 publications

(6 citation statements)

References 58 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Lite-High-Resolution Network (Lite-HRNet) [14] can rapidly estimate feature points, thereby reducing the computational complexity of the model. Feature-Selection High-Resolution network (FSHRNet) [15] adopts HRNet as the backbone and introduces a Feature Selection Convolution (FSConv) layer to fuse multi-resolution features, enabling adaptive feature selection based on object characteristics. The Improved U-Net (IU-Net) [16] enhances the HRNetv2 [17] by incorporating the csAG module, composed of spatial attention and channel attention, to improve model performance.…”

Section: Multi-scale Feature Learningmentioning

confidence: 99%

MVT: Multi-Vision Transformer for Event-Based Small Target Detection

Jing,

Lv,

Zhao

et al. 2024

Remote Sensing

View full text Add to dashboard Cite

Object detection in remote sensing plays a crucial role in various ground identification tasks. However, due to the limited feature information contained within small targets, which are more susceptible to being buried by complex backgrounds, especially in extreme environments (e.g., low-light, motion-blur scenes). Meanwhile, event cameras offer a unique paradigm with high temporal resolution and wide dynamic range for object detection. These advantages enable event cameras without being limited by the intensity of light, to perform better in challenging conditions compared to traditional cameras. In this work, we introduce the Multi-Vision Transformer (MVT), which comprises three efficiently designed components: the downsampling module, the Channel Spatial Attention (CSA) module, and the Global Spatial Attention (GSA) module. This architecture simultaneously considers short-term and long-term dependencies in semantic information, resulting in improved performance for small object detection. Additionally, we propose Cross Deformable Attention (CDA), which progressively fuses high-level and low-level features instead of considering all scales at each layer, thereby reducing the computational complexity of multi-scale features. Nevertheless, due to the scarcity of event camera remote sensing datasets, we provide the Event Object Detection (EOD) dataset, which is the first dataset that includes various extreme scenarios specifically introduced for remote sensing using event cameras. Moreover, we conducted experiments on the EOD dataset and two typical unmanned aerial vehicle remote sensing datasets (VisDrone2019 and UAVDT Dataset). The comprehensive results demonstrate that the proposed MVT-Net achieves a promising and competitive performance.

show abstract

Section: Multi-scale Feature Learningmentioning

confidence: 99%

MVT: Multi-Vision Transformer for Event-Based Small Target Detection

Jing,

Lv,

Zhao

et al. 2024

Remote Sensing

View full text Add to dashboard Cite

show abstract

“…To implement remote sensing image segmentation, ref. [12] introduced HRNet as the backbone, and a feature-selection convolution (FSConv) layer was proposed for fusing the multi-resolution features, allowing adaptive feature selection based on object features. Ref.…”

Section: Multi-resolution Feature Learningmentioning

confidence: 99%

“…The proposed HRNET in [10] can compensate for the feature loss by maintaining the high-resolution feature representation in the forward propagation. HRNet is a representative method that is applied to tasks with complex information, such as general object detection [11], image segmentation [12], object tracking [13], etc. Differently from U-Net and FPN, HRNet utilizes parallel multi-resolution sub-networks for multi-scale repeated fusion.…”

Section: Introductionmentioning

confidence: 99%

High-Resolution Network with Transformer Embedding Parallel Detection for Small Object Detection in Optical Remote Sensing Images

Zhang,

Liu,

Chang

et al. 2023

Remote Sensing

View full text Add to dashboard Cite

Small object detection in remote sensing enables the identification and analysis of unapparent but important information, playing a crucial role in various ground monitoring tasks. Due to the small size, the available feature information contained in small objects is very limited, making them more easily buried by the complex background. As one of the research hotspots in remote sensing, although many breakthroughs have been made, there still exist two significant shortcomings for the existing approaches: first, the down-sampling operation commonly used for feature extraction can barely preserve weak features of objects in a tiny size; second, the convolutional neural network methods have limitations in modeling global context to address cluttered backgrounds. To tackle these issues, a high-resolution network with transformer embedding parallel detection (HRTP-Net) is proposed in this paper. A high-resolution feature fusion network (HR-FFN) is designed to solve the first problem by maintaining high spatial resolution features with enhanced semantic information. Furthermore, a Swin-transformer-based mixed attention module (STMA) is proposed to augment the object information in the transformer block by establishing a pixel-level correlation, thereby enabling global background–object modeling, which can address the second shortcoming. Finally, a parallel detection structure for remote sensing is constructed by integrating the attentional outputs of STMA with standard convolutional features. The proposed method effectively mitigates the impact of the intricate background on small objects. The comprehensive experiment results on three representative remote sensing datasets with small objects (MASATI, VEDAI and DOTA datasets) demonstrate that the proposed HRTP-Net achieves a promising and competitive performance.

show abstract

“…Early methods based on sliding windows and candidate regions are time-consuming and have a lot of redundant calculations (Davis et al, 1975;Özden and Polat, 2005;Senthilkumaran and Rajesh, 2009;Nowozin and Lampert, 2011;Ciresan et al, 2012). In recent years, there are more and more DL-based methods in RS, including U-Net methods and its variants (Yue et al, 2019;Foivos et al, 2020), multi-scale context aggregation networks (Liu et al, 2018a;Chen et al, 2020;Xu H. et al, 2022), and multi-level feature fusion networks (Dong and Chen, 2021;Li et al, 2021b). The attention mechanism that pays attention to relevant information and ignores irrelevant information is frequently adopted with its advantages (Li et al, 2021a;Li et al, 2021b;Li YC.…”

Section: Introductionmentioning

confidence: 99%

“…Xu H. et al (2022) designed the FSHRNet using strong linear separability of high-resolution features to achieve multi-scale object segmentation in VHR images Li et al (2021c). proposed a layered self-attention model with dense connections.…”

mentioning

confidence: 99%

Deep learning-based semantic segmentation of remote sensing images: a review

Shen

et al. 2023

Front. Ecol. Evol.

View full text Add to dashboard Cite

Semantic segmentation is a fundamental but challenging problem of pixel-level remote sensing (RS) data analysis. Semantic segmentation tasks based on aerial and satellite images play an important role in a wide range of applications. Recently, with the successful applications of deep learning (DL) in the computer vision (CV) field, more and more researchers have introduced and improved DL methods to the task of RS data semantic segmentation and achieved excellent results. Although there are a large number of DL methods, there remains a deficiency in the evaluation and advancement of semantic segmentation techniques for RS data. To solve the problem, this paper surveys more than 100 papers in this field in the past 5 years and elaborates in detail on the aspects of technical framework classification discussion, datasets, experimental evaluation, research challenges, and future research directions. Different from several previously published surveys, this paper first focuses on comprehensively summarizing the advantages and disadvantages of techniques and models based on the important and difficult points. This research will help beginners quickly establish research ideas and processes in this field, allowing them to focus on algorithm innovation without paying too much attention to datasets, evaluation indicators, and research frameworks.

show abstract

Feature-Selection High-Resolution Network With Hypersphere Embedding for Semantic Segmentation of VHR Remote Sensing Images

Cited by 10 publications

References 58 publications

MVT: Multi-Vision Transformer for Event-Based Small Target Detection

MVT: Multi-Vision Transformer for Event-Based Small Target Detection

High-Resolution Network with Transformer Embedding Parallel Detection for Small Object Detection in Optical Remote Sensing Images

Deep learning-based semantic segmentation of remote sensing images: a review

Contact Info

Product

Resources

About