AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

ParCNetV2: Oversized Kernel with Enhanced Attention

Ruihan Xu, Haokui Zhang, Wenze Hu, Shiliang Zhang, Xiaoyu Wang

ICCV 2023 · Architectures and efficiency › Attention and transformers · approach: Efficient and robust vision transformers · 11 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 oversized convolution (ablation)ParCNetV2-XT · ImageNet-1K valtop-1 accuracy (%): 78.9 -> 79.4Improved
add spatial bifurcate gate unit (ablation)ParCNetV2-XT · ImageNet-1K valtop-1 accuracy (%): 79.2 -> 79.4Improved
add channel bifurcate gate unit (ablation)ParCNetV2-XT · ImageNet-1K valtop-1 accuracy (%): 79.1 -> 79.4Improved
use uniform local-global convolution block in all stages (ablation)ParCNetV2-XT · ImageNet-1K valtop-1 accuracy (%): 79.2 -> 79.4Improved
replace Swin-T/ConvNeXt-T backbone with ParCNetV2-TMask R-CNN 3x · COCOAP bbox: 46.2 -> 48.9Improved
replace ConvNeXt-T backbone with ParCNetV2-TUperNet · ADE20K (multi-scale)mIoU (%): 46.7 -> 49.4Improved
replace Swin-T with ParCNetV2-TRK3288 inference · ImageNet-1K / RK3288latency (ms): 855 -> 840Improved

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