AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Inter2Former: Dynamic Hybrid Attention for Efficient High-Precision Interactive Segmentation

You Huang, Lichao Chen, Jiayi Ji, Liujuan Cao, Shengchuan Zhang, Rongrong Ji

ICCV 2025 · Architectures and efficiency › Attention and transformers · approach: Block-sparse attention kernels · 2 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
use Dynamic Hybrid Attention instead of all BSQ attention (ablation)Inter2Former · HQSeg44K 2048NoC90: 5.64 -> 4.24Improved
use Dynamic Hybrid Attention instead of all BSQ attention (ablation)Inter2Former · HQSeg44K 20485-mIoU: 90.12 -> 92.68Improved
use DHA instead of all Full Attention (ablation)Inter2Former · HQSeg44K 1024NoC90: 5.38 -> 5.46No clear effect
use DHA instead of all Full Attention (ablation)Inter2Former · HQSeg44K 2048NoC90: 4.24 -> 4.24No clear effect
use BSQA instead of VQA (ablation)Inter2Former · HQSeg44K 2048NoC90: 4.82 -> 4.24Improved
use BSQA instead of VQA (ablation)Inter2Former · HQSeg44K 1024NoC90: 5.59 -> 5.46Improved
use Dynamic Prompt Embedding instead of full prompt embedding (ablation)Inter2Former · HQSeg44K 1024NoC90: 5.44 -> 5.46No clear effect
use Dynamic Local Upsampling instead of full upsampling (ablation)Inter2Former · DAVIS 1024NoC95: 11.81 -> 12.72Hurt
use Dynamic Local Upsampling instead of full upsampling (ablation)Inter2Former · DAVIS 2048NoC95: 7.90 -> 7.82No clear effect
increase input resolution from 1024 to 2048 (ablation)Inter2Former · HQSeg44KNoC90: 5.46 -> 4.24Improved
use CPU-optimized HMoE instead of vanilla MoE (ablation)Inter2Former · CPU latencylatency: 56-85% latency reductionImproved
use SAM distillation protocol instead of COCO+LVIS training (ablation)Inter2Former · HQSeg44K 2048NoC90: 4.58 -> 4.24Improved

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