High-Resolution Remote Sensing Image Segmentation Framework Based on Attention Mechanism and Adaptive Weighting

Abstract

Semantic segmentation has been widely used in the basic task of extracting information from images. Despite this progress, there are still two challenges: (1) it is difficult for a single-size receptive field to acquire sufficiently strong representational features, and (2) the traditional encoder-decoder structure directly integrates the shallow features with the deep features. However, due to the small number of network layers that shallow features pass through, the feature representation ability is weak, and noise information will be introduced to affect the segmentation performance. In this paper, an Adaptive Multi-Scale Module (AMSM) and Adaptive Fuse Module (AFM) are proposed to solve these two problems. AMSM adopts the idea of channel and spatial attention and adaptively fuses three-channel branches by setting branching structures with different void rates, and flexibly generates weights according to the content of the image. AFM uses deep feature maps to filter shallow feature maps and obtains the weight of deep and shallow feature maps to filter noise information in shallow feature maps effectively. Based on these two symmetrical modules, we have carried out extensive experiments. On the ISPRS Vaihingen dataset, the F1-score and Overall Accuracy (OA) reached 86.79% and 88.35%, respectively.

Exploration of Internal and External Factors of Swimmers’ Performance Based on Biofluid Mechanics and Computer Simulation

“…The human body relies on aerobic metabolism [ 30 ] and anaerobic metabolism to provide the energy needed for exercise. The energy storage of the human body can be expressed by Equation (8): where K is the energy generated by metabolism, is the horizontal propulsion energy, which provides energy for swimming movement, N is the non-propulsion energy, which enables a certain mass of water to have kinetic energy (i.e., speed) and R is the energy consumed in the form of heat energy [ 31 ].…”

Dual Attention Feature Fusion and Adaptive Context for Accurate Segmentation of Very High-Resolution Remote Sensing Images

“…Meanwhile, in remote sensing, researchers [41,42] are also inspired by the development of segmentation in natural scenes. Wang et al [43] present a gated convolutional neural network to automatically select adaptive features when merging different-layer feature maps.…”

Intelligent Planning and Research on Urban Traffic Congestion

“…The intelligent traffic system (ITS), also known as the Intelligent transportation system, includes the advanced fields of science and technology (information technology [32], computer technology [33], data communication technology [34], sensor technology [35], electronic control technology [36], automatic control theory [37], computer simulation [38],Artificial Intelligence [30], etc.) that are effectively integrated into transportation [39], service control, and vehicle manufacturing, in order to strengthen the relationship between vehicles, roads, and users, thus forming a comprehensive transportation system that plays a large and all-round role.…”

“…The intelligent traffic system will collect traffic flow data through A/D conversion [27] to send to the control center for analysis and comparison, and to obtain the ratio of traffic flow [28] size comparison. The traffic light time can be controlled according to the ratio, so the system can change according to the real-time traffic flow [29], which, coupled with the role of the overpass, can greatly improve the efficiency of the intersection [30] and improve the quality of the traffic service of the city's overall road network.…”