Point Proposal Based Instance Segmentation with Rectangular Masks for Robot Picking Task

Ito, Satoshi; Kubota, Susumu

doi:10.1007/978-3-030-69535-4_39

Cited by 6 publications

(6 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…UOAIS-Net with depth performs better than the RGB-D model in OCID, as it includes much more complex textured backgrounds than OSD. of the others are from [1], [58], and the backbone for [1] is ResNet-32 [44]. UOAIS-Net outperforms all other methods and shows that it can also generalize well on the bin environment.…”

Section: Amodal Datasetsmentioning

confidence: 95%

Unseen Object Amodal Instance Segmentation via Hierarchical Occlusion Modeling

Back¹,

Lee²,

Kim³

et al. 2021

Preprint

View full text Add to dashboard Cite

Instance-aware segmentation of unseen objects is essential for a robotic system in an unstructured environment. Although previous works achieved encouraging results, they were limited to segmenting the only visible regions of unseen objects. For robotic manipulation in a cluttered scene, amodal perception is required to handle the occluded objects behind others. This paper addresses Unseen Object Amodal Instance Segmentation (UOAIS) to detect 1) visible masks, 2) amodal masks, and 3) occlusions on unseen object instances. For this, we propose a Hierarchical Occlusion Modeling (HOM) scheme designed to reason about the occlusion by assigning a hierarchy to a feature fusion and prediction order. We evaluated our method on three benchmarks (tabletop, indoors, and bin environments) and achieved state-of-the-art (SOTA) performance. Robot demos for picking up occluded objects, codes, and datasets are available at https://sites.google.com/view/uoais.

show abstract

Section: Amodal Datasetsmentioning

confidence: 95%

Unseen Object Amodal Instance Segmentation via Hierarchical Occlusion Modeling

Back¹,

Lee²,

Kim³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…However, region proposal based segmentation methods like Mask R-CNN often fail to perform well in complex scenes where objects highly overlap, which is often the case for robotic picking. [16] proposed a method for point proposal based instance segmentation for robotic picking tasks. To generate these point proposals, [16] added specific network branches, which introduce additional overhead in terms of network parameters and computation cost.…”

Section: Related Workmentioning

confidence: 99%

“…[16] proposed a method for point proposal based instance segmentation for robotic picking tasks. To generate these point proposals, [16] added specific network branches, which introduce additional overhead in terms of network parameters and computation cost.…”

Section: Related Workmentioning

confidence: 99%

“…Note that utilizing information about our predicted grasp candidates provides a clear advantage to other point proposal based instance segmentation methods, e.g. [16,35], which need an additional module (e.g. an additional neural network branch) to predict point proposals.…”

Section: Instance Segmentation Branchmentioning

confidence: 99%

“…This additional information is used to identify and pick specific objects or a class of objects, and to understand the relation between objects in a cluttered scene. Other researchers focused on class-agnostic object instance segmentation [16,39], arguing that pixel-accurate instance segmentation of unknown objects is highly important for robotic picking, as it helps to understand each object's shape and location accurately and can be used to avoid potential collisions during grasping (as shown in [27]). Although these methods provide highly accurate results for instance segmentation, they do not directly provide grasp candidates.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Depth-aware Object Segmentation and Grasp Detection for Robotic Picking Tasks

Ainetter¹,

Böhm²,

Rohit³

et al. 2021

Preprint

View full text Add to dashboard Cite

In this paper, we present a novel deep neural network architecture for joint classagnostic object segmentation and grasp detection for robotic picking tasks using a parallelplate gripper. We introduce depth-aware Coordinate Convolution (CoordConv), a method to increase accuracy for point proposal based object instance segmentation in complex scenes without adding any additional network parameters or computation complexity. Depth-aware CoordConv uses depth data to extract prior information about the location of an object to achieve highly accurate object instance segmentation. These resulting segmentation masks, combined with predicted grasp candidates, lead to a complete scene description for grasping using a parallel-plate gripper. We evaluate the accuracy of grasp detection and instance segmentation on challenging robotic picking datasets, namely Siléane and OCID_grasp, and show the benefit of joint grasp detection and segmentation on a real-world robotic picking task.

show abstract