AI Paper Atlas

Research map / Video / Video segmentation and tracking

Hierarchical Memory Matching Network for Video Object Segmentation

Hongje Seong, Seoung Wug Oh, Joon-Young Lee, Seongwon Lee, Suhyeon Lee, Euntai Kim

ICCV 2021 · Video › Video segmentation and tracking · approach: Mask propagation for unsupervised video segmentation · 135 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 kernel guided memory matching (ablation)STM · DAVIS 2017 valJ&F: 82.2 -> 83.1Improved
use kernel guided matching instead of KMN kernelization (ablation)STM · DAVIS 2017 val / YouTube-VOS 2019J&F / G: 83.3 -> 83.1 (DAVIS), 80.0 -> 80.9 (YT19)Mixed
add top-k guided fine-scale memory matching (ablation)STM · DAVIS 2017 valJ&F: 82.2 -> 83.6Improved
add top-k guided matching on top of kernel guidance (ablation)HMMN · DAVIS 2017 valJ&F: 83.1 -> 84.7Improved
select fine-scale pixels with top-k instead of random (ablation)HMMN · DAVIS 2017 valJ&F: 81.9 -> 84.7Improved
add dropout in top-k memory module (ablation)HMMN · DAVIS 2017 valJ&F: 81.5 -> 84.7Improved
use dense fine-scale memory instead of top-k (ablation)HMMN · DAVIS 2017 valJ&F: 84.7 -> 82.0Hurt
add res2 and res3 fine-scale memory (ablation)HMMN · DAVIS 2017 valJ&F: 83.1 -> 84.7Improved
use intermediate frames for fine-scale memory (ablation)HMMN · DAVIS 2017 valJ&F: 84.7 -> 84.9No clear effect
use local 7x7 window instead of global matching in kernel tracking (ablation)HMMN · DAVIS 2017 valJ&F: 84.2 -> 84.7Improved
set sigma_init to 3 instead of 1 (ablation)HMMN · DAVIS 2017 valJ&F: 83.6 -> 84.7Improved
increase inference k from 32 to 64 (ablation)HMMN · DAVIS 2017 valJ&F: 84.7 -> 84.9No clear effect

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