Pose estimation helps the robot know where it is and where it is facing.
姿态估计帮助机器人知道自己在哪里、朝向哪里。
By combining multi-view images with a neural network, the system improved 3D pose estimation accuracy under occlusion and motion blur.
通过结合多视角图像与神经网络,该系统在遮挡和运动模糊条件下提升了三维姿态估计的准确率。