首页|Researchers from Jiangsu University Provide Details of New Studies and Findings in the Area of Robotics and Automation (Ma-stereo: Real-time Stereo Matching Via Multi-scale Attention Fusion and Spatial Error-aware Refinement)
Researchers from Jiangsu University Provide Details of New Studies and Findings in the Area of Robotics and Automation (Ma-stereo: Real-time Stereo Matching Via Multi-scale Attention Fusion and Spatial Error-aware Refinement)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
By a News Reporter-Staff News Editor at Robotics & Machine Learning DailyNews Daily News-Current study results on Robotics - Ro botics and Automation have been published.According to news reporting from Zhen jiang, People's Republic of China, by NewsRx journalists, researchstated, "Ster eo matching is a fundamental task in computer vision. Real-time stereo matching has recentlyshown great potential in robotics and autonomous driving applicatio ns."Financial support for this research came from National Natural Science Foundatio n of China (NSFC).The news correspondents obtained a quote from the research from Jiangsu Universi ty, "However, theexisting cost aggregation in real-time stereo matching suffers from accuracy limitations in ill-posed regions.Furthermore, most real-time ste reo matching methods struggle to predict disparity in object details andedge ar eas, resulting in relatively blurred and lacking detailed disparity maps. To add ress these issues, wepropose a real-time stereo matching architecture called MA -Stereo, which features a multi-scale attentionfusion (MAF) module and an atten tion-based spatial error-aware refinement (ASER) module. The MAFadaptively fuse s context and geometry information through attention mechanism, effectively impr ovingcost aggregation. In addition, the ASER refines the predicted disparity ma p, fully leveraging high-frequencyinformation and spatial evidence to accuratel y predict disparities for sharp edges and thin structures."
ZhenjiangPeople's Republic of ChinaA siaRobotics and AutomationRoboticsJiangsu University
NETL
NSTL
