AI Paper Atlas

Research map / 3D vision / Depth and stereo

Deep Sky Modeling for Single Image Outdoor Lighting Estimation

Yannick Hold-Geoffroy, Akshaya Athawale, Jean-Francois Lalonde

CVPR 2019 · 3D vision › Depth and stereo · approach: Neural light field and illumination imaging · 132 citations (OpenAlex)

Check 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
Replace the analytic Hošek-Wilkie sky representation with a learned data-driven sky modelOutdoor lighting estimation from a single image · HDR panorama relighting evaluationRMSE: almost 10× improvement in RMSEImproved
Replace the analytic Hošek-Wilkie sky representation with a learned data-driven sky modelOutdoor lighting estimation from a single image · HDR panorama relighting evaluationsi-RMSE: over 60%Improved
Use the proposed method instead of Hold-Geoffroy et al.Outdoor virtual object insertion · SUN360 user studyuser preference votes: 536 (69%) vs. 244 (31%)Improved
Use a learned sky model instead of the Hošek-Wilkie parametric sky modelOutdoor lighting estimation · SUN360 and captured HDR panorama datasetssun position estimation errorImproved
Add the decoded-sky L1 loss to the latent-vector loss (ablation)Image-to-sky encoder · Outdoor sky estimationsky energy distribution detailImproved
Train sun-azimuth and sky-parameter estimation as separate models instead of one jointly trained model (ablation)Image-to-sky and image-to-azimuth estimation · Outdoor lighting estimationgeneralization performanceImproved

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