MPF6D: Masked Pyramid Fusion 6D Pose Estimation

Object pose estimation has multiple important applications, such as robotic grasping and augmented reality. We present a new method to estimate the 6D pose of objects that improves upon the accuracy of current proposals and can still be used in real-time. Our method uses RGB-D data as input to segment objects and estimate their pose. It uses a neural network with multiple heads, one head estimates the object classification and generates the mask, the second estimates the values of the translation vector and the last head estimates the values of the quaternion that represents the rotation of the object. These heads leverage a pyramid architecture used during feature extraction and feature fusion. Our method can be used in real-time with its low inference time of 0.12 seconds and has high accuracy. With this combination of fast inference and good accuracy it is possible to use our method in robotic pick and place tasks and/or augmented reality applications.