Instance Shadow Detection with A Single-Stage Detector

TL;DR

This paper introduces the task of instance shadow detection, proposing a new dataset, an evaluation metric, and a single-stage detector with relation learning and deformable mask heads, demonstrating its effectiveness in various applications.

Contribution

It formulates the novel problem of instance shadow detection, creates a new dataset, and designs an end-to-end detector with relation learning and deformable mask heads.

Findings

The proposed method outperforms baseline approaches on the new dataset.

The detector effectively learns shadow-object relations and improves mask accuracy.

Applications include light direction estimation and photo editing.

Abstract

This paper formulates a new problem, instance shadow detection, which aims to detect shadow instance and the associated object instance that cast each shadow in the input image. To approach this task, we first compile a new dataset with the masks for shadow instances, object instances, and shadow-object associations. We then design an evaluation metric for quantitative evaluation of the performance of instance shadow detection. Further, we design a single-stage detector to perform instance shadow detection in an end-to-end manner, where the bidirectional relation learning module and the deformable maskIoU head are proposed in the detector to directly learn the relation between shadow instances and object instances and to improve the accuracy of the predicted masks. Finally, we quantitatively and qualitatively evaluate our method on the benchmark dataset of instance shadow detection and show the applicability of our method on light direction estimation and photo editing.