AI Paper Atlas

Research map / Detection and segmentation / CNN features

Graspness Discovery in Clutters for Fast and Accurate Grasp Detection

Chenxi Wang, Hao-Shu Fang, Minghao Gou, Hongjie Fang, Jin Gao, Cewu Lu

ICCV 2021 · Detection and segmentation › CNN features · approach: Efficient DETR query design · 164 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add cascaded graspness model (CGM) to Fang et al.Fang et al. · GraspNet-1Billion Seen (RealSense)AP: 27.56 -> 41.46Improved
add cascaded graspness model (CGM) to Liang et al.Liang et al. · GraspNet-1Billion Seen (RealSense)AP: 25.96 -> 33.86Improved
add cascaded graspness model (CGM) to GPDGPD · GraspNet-1Billion Seen (RealSense)AP: 22.87 -> 28.16Improved
add CGM to GPD on novel objectsGPD · GraspNet-1Billion Novel (RealSense)AP: 8.24 -> 9.73Improved
add collision detection post-processingGSNet · GraspNet-1Billion SeenAP: +1.42 AP RealSense, +2.31 AP KinectImproved
sample points by FPS instead of random sampling from whole cloud (ablation)GSNet · GraspNet-1BillionAP: 46.17 -> 51.15Improved
sample seed points from graspable landscape instead of whole cloud (FPS) (ablation)GSNet · GraspNet-1BillionAP: 51.15 -> 59.70Improved
sample seed points from graspable landscape instead of whole cloud (random) (ablation)GSNet · GraspNet-1BillionAP: 46.17 -> 53.32Improved
use graspable PVS instead of surface-normal view selection (ablation)GSNet · GraspNet-1BillionAP: 55.63 -> 59.70Improved
use graspable PVS instead of top-1 score view selection (ablation)GSNet · GraspNet-1BillionAP: 58.34 -> 59.70Improved
use scene-level instead of object-level graspness landscape (ablation)GSNet · GraspNet-1BillionAP: 55.33 -> 59.70Improved
define view graspness as feasible grasp ratio instead of max grasp quality (ablation)GSNet · GraspNet-1BillionAP: 56.95 -> 59.70Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy