Inferring 3D Object Pose in RGB-D Images
This work addresses the challenge of 3D object pose estimation and scene replacement for applications in robotics or augmented reality, representing a strong incremental improvement.
The paper tackles the problem of replacing objects in RGB-D scenes with 3D models by detecting and segmenting objects, using a CNN trained on synthetic data to predict object pose, and aligning prototypical models. It achieves a 48% relative improvement in 3D detection performance over the state-of-the-art while being an order of magnitude faster.
The goal of this work is to replace objects in an RGB-D scene with corresponding 3D models from a library. We approach this problem by first detecting and segmenting object instances in the scene using the approach from Gupta et al. [13]. We use a convolutional neural network (CNN) to predict the pose of the object. This CNN is trained using pixel normals in images containing rendered synthetic objects. When tested on real data, it outperforms alternative algorithms trained on real data. We then use this coarse pose estimate along with the inferred pixel support to align a small number of prototypical models to the data, and place the model that fits the best into the scene. We observe a 48% relative improvement in performance at the task of 3D detection over the current state-of-the-art [33], while being an order of magnitude faster at the same time.