AI Paper Atlas

Research map / 3D vision / Depth and stereo

Learning to See Through Obstructions With Layered Decomposition

Yu-Lun Liu, Wei‐Sheng Lai, Ming–Hsuan Yang, Yung‐Yu Chuang, Jia‐Bin Huang

TPAMI 2021 · 3D vision › Depth and stereo · approach: Learned optical and scene flow · 20 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
predict uniform flow fields as initialization instead of zero (ablation)flow decomposition · Vimeo-90k validationvalidation loss: 0.478 -> 0.197Improved
predict uniform flow fields instead of dense flow fields (ablation)flow decomposition · Vimeo-90k validationvalidation loss: 0.236 -> 0.197Improved
use image reconstruction network instead of temporal mean filtering (ablation)fusion · Vimeo-90k validationvalidation loss: 0.652 -> 0.197Improved
use image reconstruction network instead of temporal median filtering (ablation)fusion · Vimeo-90k validationvalidation loss: 0.555 -> 0.197Improved
add pre-training with synthetic data (ablation)online-optimization-only model · Vimeo-90k validationvalidation loss: 0.468 -> 0.197Improved
add online optimization on real data (ablation)pre-trained model · Vimeo-90k validationvalidation loss: 0.283 -> 0.197Improved
use coarse-to-fine with 5 pyramid levels instead of 1 (ablation)pyramid levels · Vimeo-90k validationvalidation loss: 0.382 -> 0.283Improved
increase pyramid levels from 5 to 6 (ablation)pyramid levels · Vimeo-90k validationvalidation loss: 0.283 -> 0.281No clear effect
add the registered frame input feature (ablation)input features · Vimeo-90k validationvalidation loss: 0.416 -> 0.283Improved
add the difference map input feature (ablation)input features · Vimeo-90k validationvalidation loss: 0.288 -> 0.283Improved
use proposed realistic training data generation (ablation)training data · Controlled sequences Stone and ToyNCC (background): 0.9390 -> 0.9428Improved
use proposed realistic training data generation (ablation)training data · Controlled sequences Stone and ToyNCC (reflection): 0.5886 -> 0.6032Improved

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