atlas

Research map / Video / Video segmentation and tracking

Learning Video Stabilization Using Optical Flow

CVPR 2020 · Video › Video segmentation and tracking · approach: Learned optical flow estimation · 93 citations (OpenAlex)

Check your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's title only (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 a pretrained optical-flow stabilization network with the proposed pipeline instead of optimization-based stabilizationVideo stabilization · Per-frame runtimeruntime: 1360ms -> 570ms; 1610ms -> 570msHelped
Add frequency-domain loss during the first training phase (ablation)Stabilization network · Trainingmotion lossHelped
Add frequency-domain loss during early training, then fine-tune using motion loss alone (ablation)Stabilization network · Trainingmotion lossHelped
Remove frequency-domain loss from training (ablation)Stabilization network · Warp-field smoothnesswarp-field smoothnessHurt
Use PCA Flow to inpaint invalid optical-flow regions (ablation)Video stabilization pipeline · Videos with moving objects and occlusionsoutput distortionHelped
Remove optical-flow masking and inpainting (ablation)Video stabilization pipeline · Videos with moving objectsoutput distortionHurt
Replace raw warp fields with PCA Flow-smoothed warp fields (ablation)Video stabilization pipeline · Warp-field valid/invalid boundariesoutput artifactsHelped
Add Gaussian noise to input optical flow with σ = 3 or σ = 10 (ablation)Proposed method · NUS datasetstability: σ = 3; σ = 10Hurt

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