KAUST DepartmentVisual Computing Center (VCC)
Computer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division
Online Publication Date2020-08-17
Print Publication Date2020-08-17
Permanent link to this recordhttp://hdl.handle.net/10754/665577
MetadataShow full item record
AbstractWe present a valid polarization-based reflection contaminated image synthesis method, which can provide adequate, diverse and authentic training dataset. Meanwhile, we enhance the neural network by introducing the reflection information as guidance and utilizing adaptive convolution kernel size to fuse multi-scale information. We demonstrate that the proposed approach achieves convincing improvements over state of the arts.
CitationPang, Y., Yuan, M., Fu, Q., & Yan, D.-M. (2020). Reflection Removal via Realistic Training Data Generation. ACM SIGGRAPH 2020 Posters. doi:10.1145/3388770.3407419
SponsorsThis work was supported by the National Key R&D Program of China (2019YFB2204104 and 2018YFB2100602). (Portions of) the research in this paper used the 'SIR2' Dataset made available by the ROSE Lab at the Nanyang Technological University, Singapore.
Conference/Event nameACM SIGGRAPH 2020 Posters - International Conference on Computer Graphics and Interactive Techniques, SIGGRAPH 2020