AI Paper Atlas

Research map / 3D vision / 3D generation

SemGrasp: Semantic Grasp Generation via Language Aligned Discretization

Kailin Li, Jingbo Wang, Lixin Yang, Cewu Lu, Bo Dai

ECCV 2024 · 3D vision › 3D generation · approach: Hand-object interaction generation · 34 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 the refinement token to the grasp representation (ablation)SemGrasp VQ-VAE · OakInk reconstructionMPVPE (mm): 20.36 -> 14.97Improved
use three hierarchical tokens instead of a single token (ablation)SemGrasp VQ-VAE · OakInk reconstructionMPVPE (mm): 29.95 -> 14.97Improved
use more than one refinement token (2 or 3) (ablation)SemGrasp VQ-VAE · OakInk reconstructionSIV (cm3): 2.72 -> 2.98Hurt
use hierarchical VQ-VAE instead of a single shared VQ-VAE (ablation)SemGrasp VQ-VAE · OakInk reconstructionMPVPE (mm): 28.02 -> 14.97Improved
assign semantic meaning (separate losses) to the three tokens (ablation)SemGrasp VQ-VAE · OakInk reconstructionMPVPE (mm): 21.94 -> 14.97Improved
use 512 codebook entries instead of 256 or 1024 (ablation)SemGrasp VQ-VAE · OakInk reconstructionMPVPE (mm): 95.52 -> 14.97Improved
add EMA and codebook reset to VQ-VAE training (ablation)SemGrasp VQ-VAE · OakInk reconstructionMPVPE (mm): 14.97 -> 24.67Hurt
use Vicuna-7B instead of Llama-7B (ablation)SemGrasp MLLM · language-guided grasp generationGPT-4 score: 58.9 -> 74.5Improved
add object size token (ablation)SemGrasp MLLM · language-guided grasp generationGPT-4 score: 43.3 -> 74.5Improved
use two-stage training (alignment then instruction tuning) (ablation)SemGrasp MLLM · language-guided grasp generationGPT-4 score: 62.5 -> 74.5Improved
increase LoRA rank from 64 to 128 (ablation)SemGrasp MLLM · language-guided grasp generationGPT-4 score: 74.5 -> 74.5No clear effect
use MLLM instead of BERT classification baselinelanguage-guided grasp generation · OakInk testGPT-4 score: 47.3 -> 74.5Improved

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