RenDetNet: Weakly-supervised Shadow Detection with Shadow Caster Verification

Nikolina Kubiak, Elliot Wortman, Armin Mustafa, Graeme Phillipson, Stephen Jolly +1 more
8/30/2024
cs.CVcs.GR

No official code found. Best implementation: n-kubiak/RenDetNet (1 star, last commit 2024-09-27).

Abstract

Existing shadow detection models struggle to differentiate dark image areas from shadows. In this paper, we tackle this issue by verifying that all detected shadows are real, i.e. they have paired shadow casters. We perform this step in a physically-accurate manner by differentiably re-rendering the scene and observing the changes stemming from carving out estimated shadow casters. Thanks to this approach, the RenDetNet proposed in this paper is the first learning-based shadow detection model whose supervisory signals can be computed in a self-supervised manner. The developed system compares favourably against recent models trained on our data. As part of this publication, we release our code on github.

View on arXivView PDF

Code Implementations(1)

n-kubiak/RenDetNetCommunity100%

Official repo for the RenDetNet project - AIM @ ECCV2024

10Aug 23, 20242 years ago

Similar Papers

Physics-Grounded Attached Shadow Detection Using Approximate 3D Geometry and Light Direction

86% match1 repo12025

SILT: Shadow-aware Iterative Label Tuning for Learning to Detect Shadows from Noisy Labels

86% match1 repo162026

DTTNet: Improving Video Shadow Detection via Dark-Aware Guidance and Tokenized Temporal Modeling

86% match1 repo2025

Test-Time Intensity Consistency Adaptation for Shadow Detection

85% match1 repo72024

Unveiling Deep Shadows: A Survey and Benchmark on Image and Video Shadow Detection, Removal, and Generation in the Deep Learning Era

85% match1 repo1102024

Adaptive Illumination Mapping for Shadow Detection in Raw Images

83% match2026

Cite this paper

@article{kubiak2024rendetnet,
  title  = {RenDetNet: Weakly-supervised Shadow Detection with Shadow Caster Verification},
  author = {Nikolina Kubiak and Elliot Wortman and Armin Mustafa and Graeme Phillipson and Stephen Jolly and Simon Hadfield},
  year   = {2024},
  eprint = {2408.17143},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2408.17143v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2408.17143.svg)](https://codeofpaper.com/paper/2408.17143)

Discussion