RenDetNet: Weakly-supervised Shadow Detection with Shadow Caster Verification
This addresses shadow detection for computer vision applications, offering a novel self-supervised approach that is incremental in improving accuracy.
The paper tackles the problem of distinguishing shadows from dark image areas by verifying detected shadows have paired shadow casters using differentiable re-rendering, resulting in a self-supervised learning-based model that outperforms recent models on their data.
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.