mlsanityData to 13 Oct 2026

PhaSR: Generalized Image Shadow Removal with Physically Aligned Priors

arXiv 2601.17470 ↗First posted 24 Jan 2026cs.CVcited by 4

Not on the map. Why?

Abstract (arXiv)

Shadow removal under diverse lighting conditions requires disentangling illumination from intrinsic reflectance, a challenge compounded when physical priors are not properly aligned. We propose PhaSR (Physically Aligned Shadow Removal), addressing this through dual-level prior alignment to enable robust performance from single-light shadows to multi-source ambient lighting. First, Physically Aligned Normalization (PAN) performs closed-form illumination correction via Gray-world normalization, log-domain Retinex decomposition, and dynamic range recombination, suppressing chromatic bias. Second, Geometric-Semantic Rectification Attention (GSRA) extends differential attention to cross-modal alignment, harmonizing depth-derived geometry with DINO-v2 semantic embeddings to resolve modal conflicts under varying illumination. Experiments show competitive performance in shadow removal with lower complexity and generalization to ambient lighting where traditional methods fail under multi-source illumination. Our source code is available at https://github.com/ming053l/PhaSR.

Key facts. Cited by 4 papers in the map (data through 13 Oct 2026). Builds on 32 of its 57 references that are in the map. First posted on arXiv 24 Jan 2026.

Builds on

32 of 57 references.

References · 32

  1. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,656
  2. 2023Segment Anythingcited by 8,195
  3. 2024Depth Anything V2cited by 1,399
  4. 2020Learning Enriched Features for Real Image Restoration and Enhancementcited by 311
  5. 2017Stacked Conditional Generative Adversarial Networks for Jointly Learning Shadow Detection and Shadow Removalcited by 161
  6. 2019Towards Ghost-free Shadow Removal via Dual Hierarchical Aggregation Network and Shadow Matting GANcited by 103
  7. 2022ShadowDiffusion: When Degradation Prior Meets Diffusion Model for Shadow Removalcited by 103
  8. 2019Shadow Removal via Shadow Image Decompositioncited by 97
  9. 2021Auto-Exposure Fusion for Single-Image Shadow Removalcited by 79
  10. 2022DC-ShadowNet: Single-Image Hard and Soft Shadow Removal Using Unsupervised Domain-Classifier Guided Networkcited by 69
  11. 2022Removing Objects From Neural Radiance Fieldscited by 66
  12. 2020From Shadow Segmentation to Shadow Removalcited by 59
20 more
  1. 2018Direction-aware Spatial Context Features for Shadow Detection and Removalcited by 58
  2. 2017Direction-aware Spatial Context Features for Shadow Detectioncited by 56
  3. 2022DeS3: Adaptive Attention-driven Self and Soft Shadow Removal using ViT Similaritycited by 38
  4. 2025UniRelight: Learning Joint Decomposition and Synthesis for Video Relightingcited by 38
  5. 2023Latent Feature-Guided Diffusion Models for Shadow Removalcited by 29
  6. 2024Recasting Regional Lighting for Shadow Removalcited by 23
  7. 2024ShadowRefiner: Towards Mask-free Shadow Removal via Fast Fourier Transformercited by 20
  8. 2023SIRe-IR: Inverse Rendering for BRDF Reconstruction with Shadow and Illumination Removal in High-Illuminance Scenescited by 19
  9. 2024OmniSR: Shadow Removal under Direct and Indirect Lightingcited by 19
  10. 2025LIRM: Large Inverse Rendering Model for Progressive Reconstruction of Shape, Materials and View-dependent Radiance Fieldscited by 16
  11. 2024Gaussian Shadow Casting for Neural Characterscited by 14
  12. 2024Detail-Preserving Latent Diffusion for Stable Shadow Removalcited by 13
  13. 2024Regional Attention for Shadow Removalcited by 12
  14. 2024Towards Image Ambient Lighting Normalizationcited by 9
  15. 2025DenseSR: Image Shadow Removal as Dense Predictioncited by 5
  16. 2024ShadowHack: Hacking Shadows via Luminance-Color Divide and Conquercited by 4
  17. 2025Retinex-guided Histogram Transformer for Mask-free Shadow Removalcited by 4
  18. 2025After the Party: Navigating the Mapping From Color to Ambient Lightingcited by 4
  19. 2024Real-Time Compressed Sensing for Joint Hyperspectral Image Transmission and Restoration for CubeSatcited by 3
  20. 2025Neural Inverse Rendering from Propagating Lightcited by 1

Built on by

Cited by year (2026 to 13 Oct)

Citing papers per year: 2026: 4.

4 papers.

  1. 2026CANDLE: Illumination-Invariant Semantic Priors for Color Ambient Lighting Normalizationcited by 0
  2. 2026CFSR: Geometry-Conditioned Shadow Removal via Physical Disentanglementcited by 0
  3. 2026WildShadowRemover: In-the-Wild Video Shadow Removal via Detail-Preserving Video Diffusion Modelscited by 0
  4. 2026After a Decade: Bringing Shadow Removal into the Real World with Agentic Training Datacited by 0