CPNet: A Context Preserver Convolutional Neural Network for Detecting Shadows in Single RGB Images

TL;DR

This paper introduces CPNet, a novel deep learning CNN architecture that detects shadows in single RGB images by preserving context and capturing global and local shadow patterns, significantly improving accuracy.

Contribution

The paper presents a new CNN architecture that maintains context during training for pixel-level shadow detection in single RGB images, outperforming previous methods.

Findings

Improved BER by 22% on SBU dataset.

Enhanced BER by 14% on UCF dataset.

Effective global and local shadow pattern detection.

Abstract

Automatic detection of shadow regions in an image is a difficult task due to the lack of prior information about the illumination source and the dynamic of the scene objects. To address this problem, in this paper, a deep-learning based segmentation method is proposed that identifies shadow regions at the pixel-level in a single RGB image. We exploit a novel Convolutional Neural Network (CNN) architecture to identify and extract shadow features in an end-to-end manner. This network preserves learned contexts during the training and observes the entire image to detect global and local shadow patterns simultaneously. The proposed method is evaluated on two publicly available datasets of SBU and UCF. We have improved the state-of-the-art Balanced Error Rate (BER) on these datasets by 22\% and 14\%, respectively.