AI Paper Atlas

Research map / Video / Action recognition

Omni-sourced Webly-supervised Learning for Video Recognition

Haodong Duan, Yue Zhao, Yuanjun Xiong, Wentao Liu, Dahua Lin

ECCV 2020 · Video › Action recognition · approach: Contrastive video representation learning · 100 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 OmniSource web dataTSN-3seg ResNet50 ImageNet · Kinetics-400 valTop-1 accuracy: 70.6 -> 73.6Improved
add OmniSource web dataSlowOnly-4x16 R50 from scratch · Kinetics-400 valTop-1 accuracy: 72.9 -> 76.8Improved
add OmniSource web dataSlowOnly-4x16 R50 ImageNet · Kinetics-400 valTop-1 accuracy: 73.8 -> 76.6Improved
add OmniSource web datairCSN-152 IG-65M · Kinetics-400 valTop-1 accuracy: 82.6 -> 83.6Improved
add OmniSource web dataSlowOnly-8x8 R101 from scratch · Kinetics-400 valTop-1 accuracy: 76.3 -> 80.4Improved
skip teacher filtering of web data (ablation)TSN joint training · Kinetics-400 valTop-1 accuracy: 70.6 -> 67.4Hurt
add Google and Instagram images with teacher filtering (ablation)TSN-3seg R50 · Kinetics-400 valTop-1 accuracy: 70.6 -> 72.0Improved
add untrimmed videos (ablation)TSN-3seg R50 · Kinetics-400 valTop-1 accuracy: 70.6 -> 71.7Improved
use all sources combined (ablation)TSN-3seg R50 · Kinetics-400 valTop-1 accuracy: 70.6 -> 73.6Improved
use class-agnostic perspective warp instead of still replication (ablation)SlowOnly-4x16 R50 · Kinetics-400 valTop-1 accuracy: 74.1 -> 74.5Improved
use random translation to turn images into clips (ablation)SlowOnly-4x16 R50 · Kinetics-400 valTop-1 accuracy: 73.8 -> 73.7Hurt
add mixup (ablation)SlowOnly-4x16 R50 from scratch · Kinetics-400 valTop-1 accuracy: 72.9 -> 73.3Improved

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