Deep Learning of Local RGB-D Patches for 3D Object Detection and 6D Pose Estimation

Abstract

We present a 3D object detection method that uses regressed descriptors of locally-sampled RGB-D patches for 6D vote casting. For regression, we employ a convolutional auto-encoder that has been trained on a large collection of random local patches. During testing, scene patch descriptors are matched against a database of synthetic model view patches and cast 6D object votes which are subsequently filtered to refined hypotheses. We evaluate on three datasets to show that our method generalizes well to previously unseen input data, delivers robust detection results that compete with and surpass the state-of-the-art while being scalable in the number of objects.

arXiv commentTo appear at ECCV 2016

Cited by in corpus (11)

• PoseCNN: A Convolutional Neural Network for 6D Object Pose Estimation in Cluttered Scenes
• Rotational Subgroup Voting and Pose Clustering for Robust 3D Object Recognition
• PPFNet: Global Context Aware Local Features for Robust 3D Point Matching
• A Deep Learning Approach for Pose Estimation from Volumetric OCT Data
• Estimating 6D Pose From Localizing Designated Surface Keypoints
• Multi-Task Deep Networks for Depth-Based 6D Object Pose and Joint Registration in Crowd Scenarios
• 6D Object Pose Estimation with Depth Images: A Seamless Approach for Robotic Interaction and Augmented Reality
• ApolloCar3D: A Large 3D Car Instance Understanding Benchmark for Autonomous Driving
• 6D Object Pose Estimation without PnP
• Visual-Inertial-Semantic Scene Representation for 3-D Object Detection
• Category-level 6D Object Pose Recovery in Depth Images