Robust Instance Recognition in Presence of Occlusion and Clutter

Abstract

Abstract. We present a robust learning based instance recognition framework from single view point clouds. Our framework is able to handle real-world instance recognition challenges, i.e, clutter, similar looking distractors and occlusion. Recent algorithms have separately tried to address the problem of clutter [9] and occlusion [16] but fail when these challenges are combined. In comparison we handle all challenges within a single framework. Our framework uses a soft label Random Forest [5] to learn discriminative shape features of an object and use them to classify both its location and pose. We propose a novel iterative training scheme for forests which maximizes the margin between classes to improve recognition accuracy, as compared to a conventional training procedure. The learnt forest outperforms template matching, DPM [7] in presence of similar looking distractors. Using occlusion information, computed from the depth data, the forest learns to emphasize the shape features from the visible regions thus making it robust to occlusion. We benchmark our system with the state-of-the-art recognition systems [9,7] in challenging scenes drawn from the largest publicly available dataset. To complement the lack of occlusion tests in this dataset, we introduce our Desk3D dataset and demonstrate that our algorithm outperforms other methods in all settings.

Large-scale 6D Object Pose Estimation Dataset for Industrial Bin-Picking

“…Datasets sharing similar properties were presented in [12], [13], [6], [14]. All of these datasets have limited pose variability and data redundancy since only the very same scene is recorded from different angles (see Table I).…”

Symmetry Aware Evaluation of 3D Object Detection and Pose Estimation in Scenes of Many Parts in Bulk

“…Several approaches have been proposed to evaluate 3D object detection and pose estimation for manipulation tasks based on robotic experiments [7,8,9,10], but pose accuracy is difficult to evaluate in such scenario due to the lack Figure 1: Samples exhibiting the strong correlation of data within existing datasets. From left to right: datasets of Tejani et al [5], Hinterstoisser et al [4,13], T-LESS [16] and Desk3D [14]. of ground truth, and its online nature makes reproducibility difficult.…”

What, Where and How? Introducing pose manifolds for industrial object manipulation

“…More recently, Brachmann et al (2014) introduced a new representation in form of a joint 3D object coordinate and class labeling, which, however, suffers in cases of occlusions. Song and Xiao (2014) proposed a computationally expensive approach to the 6 DoF pose estimation problem that slides exemplar SVMs in the 3D space, while in Bonde et al (2014) shape priors are learnt by soft labeling random forest for 3D object classification and pose estimation. In turn, part-based approaches focus on learning distinctive object models from wide training collections to strive the partial occlusion challenge.…”

“…Our main contributions can be summarized as follows: Compared to the state of the art works in object recognition and pose estimation (Brachmann et al, 2014;Bonde, Badrinarayanan, & Cipolla, 2014;Hinterstoisser et al, 2011;Lim, Khosla, & Torralba, 2014;Tejani, Tang, Kouskouridas, & Kim, 2014;Wohlhart & Lepetit, 2015) our method offers higher generalization capabilities through the recognition of objects that do not have to belong in the training dataset. Additionally, our sophisticated manifold modeling technique builds compact and object-class invariant manifolds that are not prone to occlusions.…”