Research map / 3D vision / Depth and stereo
Learning Intrinsic Image Decomposition From Watching the World
Read the paperCheck 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 changed | Where | Measured | Effect |
|---|---|---|---|
| add all-pairs reconstruction loss rather than per-image reconstruction loss (ablation) | BigTime-trained intrinsic image model · IIW test | WHDR%: 25.9 -> 20.3 | Helped |
| add dense spatio-temporal reflectance smoothness rather than local reflectance smoothness (ablation) | BigTime-trained intrinsic image model · IIW test | WHDR%: 27.4 -> 20.3 | Helped |
| use colored shading rather than grayscale shading (ablation) | BigTime-trained intrinsic image model · IIW test | WHDR%: 22.3 -> 20.3 | Helped |
| add the shading-smoothness component omitted in Eq. [ref] (ablation) | BigTime-trained intrinsic image model · SAW test | AP%: 97.15 -> 97.90 | Helped |
| use the full shading smoothness rather than local shading smoothness (ablation) | BigTime-trained intrinsic image model · SAW test | AP%: 97.03 -> 97.90 | Helped |
| train on unlabeled time-lapse sequences rather than supervised ground-truth decompositions | intrinsic image decomposition · MIT intrinsic test | MSE avg.: 0.0141 vs. 0.0115 | Mixed |
| train on BigTime sequences without IIW annotations | intrinsic image decomposition · IIW test | WHDR%: 25.9 -> 20.3 | Helped |
Closest papers by meaning
- IDArb: Intrinsic Decomposition for Arbitrary Number of Input Views and Illuminations (ICLR 2025 · similarity 0.86)
- Intrinsic Image Decomposition for Robust Self-supervised Monocular Depth Estimation on Reflective Surfaces (AAAI 2025 · similarity 0.86)
- When Shadow Removal Meets Intrinsic Image Decomposition: A Joint Learning Framework Using Unpaired Data (AAAI 2025 · similarity 0.86)
- ReasonX: MLLM-Guided Intrinsic Image Decomposition (CVPR 2026 · similarity 0.85)
- Intrinsic Image Diffusion for Indoor Single-view Material Estimation (CVPR 2024 · similarity 0.85)
- Latent Intrinsics Emerge from Training to Relight (NeurIPS 2024 · similarity 0.85)
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.