首页|Investigators at Shanghai Jiao Tong University Report Findings in Robotics and A utomation (Fgct6d: Frequency-guided Cnntransformer Fusion Network for Metal Par ts’ Robust 6d Pose Estimation)
Investigators at Shanghai Jiao Tong University Report Findings in Robotics and A utomation (Fgct6d: Frequency-guided Cnntransformer Fusion Network for Metal Par ts’ Robust 6d Pose Estimation)
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 Daily News Daily News – Data detailed on Robotics - Robotics a nd Automation have been presented. According to news reporting out of Shanghai, People’s Republic of China, by NewsRx editors, research stated, “The 6D pose est imation for metal parts is essential in industrial robotic applications. The col or homogeneity, texture-less and light-reflecting properties of metal parts rais e great challenges.” Our news journalists obtained a quote from the research from Shanghai Jiao Tong University, “Current 6D pose estimation methods have gained extensive concern us ing CNNs. However, these CNN-based methods lack Transformer’s ability to focus o n extracting low-frequency features and long-range context information. In the l etter, we explore taking full advantage of CNN and Transformer from a frequencydomain perspective to enhance the performance of metal parts’ 6D pose estimation . Specifically, we propose a frequency-guided CNN-Transformer fusion 6D pose est imation network (FGCT6D). First, we construct a novel pixel attention residual m odule to improve the high-frequency attention of CNN. Then, we design a dual-bra nch CNN-Transformer encoder: the Swin-Transformer extracts global information an d low-frequency features, and the CNN captures local information and high-freque ncy features. Second, the frequency-guided feature fusion module is proposed to fuse the extracted multi-spectral features. Third, to maximize the utilization o f the rich frequency-domain feature representation, we propose a feature fusion decoder with Conv-MSA modules. Additionally, we leverage optimal transport theor y, treating dense correspondences as spatial probability distributions, and desi gn the optimal transport loss function.”
ShanghaiPeople’s Republic of ChinaAs iaRobotics and AutomationRoboticsShanghai Jiao Tong University
NETL
NSTL
