AI Paper Atlas

Research map / Video / Video segmentation and tracking

Time Lens: Event-Based Video Frame Interpolation

CVPR 2021 · Video › Video segmentation and tracking · approach: Event-based video reconstruction · 227 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
Combine event-derived warping, synthesis, refinement, and attention averagingTime Lens · Middlebury, Vimeo90k, GoPro, HQF, and HS-ERGBPSNR: up to 5.21 dB improvementHelped
Add attention averaging to combine synthesis and warping-based outputs (ablation)Time Lens · Vimeo90k (denoising) validationPSNR: 34.10 -> 35.83Helped
Add warping refinement after warping interpolation (ablation)Time Lens · Vimeo90k (denoising) validationPSNR: 26.68 -> 33.02Helped
Remove access to inter-frame events (ablation)Time Lens · Middlebury, 1 frame skipPSNR: 33.27 -> 29.03Hurt
Fine-tune on real event data (ablation)Time Lens · HQFPSNR: 30.57 -> 32.49Helped
Use event-derived optical flow for warping rather than relying on frame-derived flowTime Lens · Highly dynamic interpolationPSNRHelped
Use synthesis rather than warping in regions with lighting changes or newly appearing objectsTime Lens · Highly dynamic interpolationPSNRMixed
Use warping rather than synthesis when event data is noisy or insufficientTime Lens · Highly dynamic interpolationPSNRMixed

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