Segmenting Unknown 3D Objects from Real Depth Images using Mask R-CNN Trained on Synthetic Point Clouds

Abstract

The ability to segment unknown objects in depth images has potential to enhance robot skills in grasping and object tracking. Recent computer vision research has demonstrated that Mask R-CNN can be trained to segment specific categories of objects in RGB images when massive hand labeled datasets are available. As generating these datasets is time-consuming, we instead train with synthetic depth images. Many robots now use depth sensors, and recent results suggest training on synthetic depth data can generalize well to the real world. We present a method for automated dataset generation and rapidly generate a training dataset of 50k depth images and 320k object masks synthetically using simulated scenes of 3D CAD models. We train a variant of Mask R-CNN on the generated dataset to perform category-agnostic instance segmentation without hand-labeled data. We evaluate the trained network, which we refer to as Synthetic Depth (SD) Mask R-CNN, on a set of real, high-resolution images of challenging, densely cluttered bins containing objects with highly-varied geometry. SD Mask R-CNN outperforms point cloud clustering baselines by an absolute 15% in Average Precision and 20% in Average Recall, and achieves performance levels similar to a Mask RCNN trained on a massive, hand-labeled RGB dataset and fine-tuned on real images from the experimental setup. The network also generalizes well to a lower-resolution depth sensor. We deploy the model in an instance-specific grasping pipeline to demonstrate its usefulness in a robotics application.