TY - GEN
T1 - Reflection Removal via Realistic Training Data Generation
AU - Pang, Youxin
AU - Yuan, Mengke
AU - Fu, Qiang
AU - Yan, Dong Ming
N1 - KAUST Repository Item: Exported on 2020-10-15
Acknowledgements: This 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.
PY - 2020/8/15
Y1 - 2020/8/15
N2 - We 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.
AB - We 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.
UR - http://hdl.handle.net/10754/665577
UR - https://dl.acm.org/doi/10.1145/3388770.3407419
UR - http://www.scopus.com/inward/record.url?scp=85091969870&partnerID=8YFLogxK
U2 - 10.1145/3388770.3407419
DO - 10.1145/3388770.3407419
M3 - Conference contribution
SN - 9781450379731
BT - ACM SIGGRAPH 2020 Posters
PB - ACM
ER -