AI Paper Atlas

Research map / Video / Action recognition

K-Centered Patch Sampling for Efficient Video Recognition

Seong Hyeon Park, Jihoon Tack, Byeongho Heo, Jung-Woo Ha, Jinwoo Shin

ECCV 2022 · Video › Action recognition · approach: Contrastive video representation learning · 17 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
sample random patches instead of random framesDeiT-base video extension · Something-Something v2Top-1 accuracy: 58.95 -> 61.64Improved
use K-centered sampling instead of random patch samplingDeiT-base video extension · Something-Something v2Top-1 accuracy: 61.64 -> 64.31Improved
use K-centered patch sampling instead of frame-basedMotionformer · Kinetics-400Top-1 accuracy: 71.42 -> 74.85Improved
use K-centered patch sampling instead of frame-basedViT · Something-Something v2Top-1 accuracy: 62.50 -> 63.83Improved
use K-centered patch sampling instead of frame-basedXViT · Something-Something v2Top-1 accuracy: 62.40 -> 62.78Improved
use K-centered patch sampling instead of frame-basedTimeSformer · Kinetics-400Top-1 accuracy: 77.95 -> 77.98No clear effect
use K-centered patch sampling instead of frame-basedTimeSformer · Something-Something v2Top-1 accuracy: 63.76 -> 63.81No clear effect
choose patches using RGB instead of grayscale distance (ablation)ViT with K-centered sampling · Kinetics-200Top-1 accuracy: 81.94 -> 82.90Improved
add 3 complete frames via hybrid sampling (ablation)ViT with K-centered sampling · Kinetics-200Top-1 accuracy: 81.56 -> 82.90Improved
add hybrid frames to patch sampling (ablation)Motionformer with K-centered sampling · Kinetics-200Top-1 accuracyNo clear effect
raise distance coefficients from ws=1, wt=1 to ws=10, wt=10 (ablation)ViT with K-centered sampling · Kinetics-200Top-1 accuracy: 82.90 -> 81.56Hurt
train with K-centered sampling instead of frame-based, test with 50% patches dropped (ablation)ViT · Kinetics-200Top-1 accuracy: 75.64 -> 79.28Improved

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