Revisiting Shadow Detection from a Vision-Language Perspective

• URL: http://arxiv.org/abs/2605.11771v1
• Date: Tue, 12 May 2026 08:42:30 GMT
• Title: Revisiting Shadow Detection from a Vision-Language Perspective
• Abstract summary: Shadow Vision--Language (SVL) uses language as an explicit semantic reference to disambiguate shadows from visually similar dark regions.<n>SVL applies local patch-level constraints with text embeddings to improve fine-grained discrimination under challenging appearance conditions.<n>Built on a frozen DINOv3 image encoder, the framework learns only lightweight projection and decoding modules, yielding a parameter-efficient design with less than $1%$ trainable parameters.
• Score: 88.46763399969217
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Shadow detection is commonly formulated as a vision-driven dense prediction problem, where models rely primarily on pixel-wise visual supervision to distinguish shadows from non-shadow regions. However, this formulation can become unreliable in visually ambiguous cases, where similar dark regions may correspond either to cast shadows or to intrinsically dark surfaces, making visual evidence alone insufficient for establishing a stable decision rule. In this work, we revisit shadow detection from a vision--language perspective and argue that robust prediction benefits from an explicit semantic reference beyond visual cues alone. We propose SVL, a Shadow Vision--Language framework that uses language as an explicit semantic reference to disambiguate shadows from visually similar dark regions. SVL aligns the global image representation with shadow-related text embeddings through a scene-level shadow ratio regression objective, thereby providing image-level guidance on the overall extent of shadows. To transfer this global guidance to dense inference, SVL introduces a global-to-local coupling mechanism that enforces consistency between image-level guidance and patch-level predictions. In parallel, SVL applies local patch-level constraints with text embeddings to improve fine-grained discrimination under challenging appearance conditions. Built on a frozen DINOv3 image encoder, the framework learns only lightweight projection and decoding modules, yielding a parameter-efficient design with less than $1\%$ trainable parameters. Extensive experiments on multiple shadow detection benchmarks, including dedicated hard-case evaluations, suggest strong overall performance and improved robustness under visually ambiguous conditions.

Related papers

• Consistency as Regularization for Unsupervised Shadow Removal [1.2289361708127875]
  We propose ShadowCLR, an unsupervised framework that learns shadow removal directly from shadow images.<n>Our key observation is that shadows vary across observations while the underlying scene content remains largely consistent.<n>Experiments on multiple benchmarks show that ShadowCLR competitive and often superior performance over state-of-the-art unsupervised methods.
  arXiv  Detail & Related papers  (2026-09-01T19:26:25Z)
• VSDiffusion: Taming Ill-Posed Shadow Generation via Visibility-Constrained Diffusion [10.547624877131122]
  We propose VSDiffusion to generate realistic cast shadows for inserted foreground objects.<n>A visibility-constrained two-stage framework designed to narrow the solution space by incorporating visibility priors.<n> Experiments on widely used public DESOBAv2 dataset demonstrated that our proposed VSDiffusion can generate accurate shadow.
  arXiv  Detail & Related papers  (2026-03-09T06:52:53Z)
• DeshadowMamba: Deshadowing as 1D Sequential Similarity [85.07259906446588]
  We introduce Mamba, a selective state space model that propagates global context through directional state transitions.<n>Despite its potential, directly applying Mamba to image data is suboptimal, since it lacks awareness of shadow-non-shadow semantics.<n>We propose CrossGate, a directional modulation mechanism that injects shadow-aware similarity into Mamba's input gate.<n>To further ensure appearance fidelity, we introduce ColorShift regularization, a contrastive learning objective driven by global color statistics.
  arXiv  Detail & Related papers  (2025-10-28T10:14:23Z)
• Test-Time Intensity Consistency Adaptation for Shadow Detection [35.03354405371279]
  TICA is a novel framework that leverages light-intensity information during test-time adaptation to enhance shadow detection accuracy. A basic encoder-decoder model is initially trained on a labeled dataset for shadow detection. During the testing phase, the network is adjusted for each test sample by enforcing consistent intensity predictions.
  arXiv  Detail & Related papers  (2024-10-10T08:08:32Z)
• SwinShadow: Shifted Window for Ambiguous Adjacent Shadow Detection [90.4751446041017]
  We present SwinShadow, a transformer-based architecture that fully utilizes the powerful shifted window mechanism for detecting adjacent shadows. The whole process can be divided into three parts: encoder, decoder, and feature integration. Experiments on three shadow detection benchmark datasets, SBU, UCF, and ISTD, demonstrate that our network achieves good performance in terms of balance error rate (BER)
  arXiv  Detail & Related papers  (2024-08-07T03:16:33Z)
• Delving into Dark Regions for Robust Shadow Detection [47.60700654394781]
  State-of-the-art deep methods tend to have higher error rates in differentiating shadow pixels from non-shadow pixels in dark regions. We propose a novel shadow detection approach that first learns global contextual cues over the entire image and then zooms into the dark regions to learn local shadow representations.
  arXiv  Detail & Related papers  (2024-02-21T09:07:07Z)
• Towards Better Gradient Consistency for Neural Signed Distance Functions via Level Set Alignment [50.892158511845466]
  We show that gradient consistency in the field, indicated by the parallelism of level sets, is the key factor affecting the inference accuracy. We propose a level set alignment loss to evaluate the parallelism of level sets, which can be minimized to achieve better gradient consistency.
  arXiv  Detail & Related papers  (2023-05-19T11:28:05Z)
• ShadowFormer: Global Context Helps Image Shadow Removal [41.742799378751364]
  It is still challenging for the deep shadow removal model to exploit the global contextual correlation between shadow and non-shadow regions. We first propose a Retinex-based shadow model, from which we derive a novel transformer-based network, dubbed ShandowFormer. A multi-scale channel attention framework is employed to hierarchically capture the global information. We propose a Shadow-Interaction Module (SIM) with Shadow-Interaction Attention (SIA) in the bottleneck stage to effectively model the context correlation between shadow and non-shadow regions.
  arXiv  Detail & Related papers  (2023-02-03T10:54:52Z)
• Estimating Reflectance Layer from A Single Image: Integrating Reflectance Guidance and Shadow/Specular Aware Learning [66.36104525390316]
  We propose a two-stage learning method, including reflectance guidance and a Shadow/Specular-Aware (S-Aware) network to tackle the problem. In the first stage, an initial reflectance layer free from shadows and specularities is obtained with the constraint of novel losses. To further enforce the reflectance layer to be independent of shadows and specularities in the second-stage refinement, we introduce an S-Aware network that distinguishes the reflectance image from the input image.
  arXiv  Detail & Related papers  (2022-11-27T07:26:41Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.