Abstract
Recently, shadow detectors based on Deep Convolutional Neural Networks (DCNN) have lifted the detection performances to a new height. However, the correct detection of mild and very dark shadows remains to be a difficult task. In this work, we propose a new model called Double-stream Atrous Network (DSAN) for shadow detection. It has a double-streams framework: the pooling stream extracts high-level features and the residual stream incorporates low-level features with high-level feature maps from the pooling stream in each single step. We also design new modules such as Atrous Convolution Module (ACM), Multi-layer Atrous Pooling Module (MLAPM), and Cross-stream Residual Module (CSRM) to extract shadow features effectively. On two shadow datasets, our DSAN outperforms several popular shadow detectors based on DCNN.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
