Add to BibSonomyAbstract
Conventional methods feed multi-modal images indiscriminately into model during training, introducing possible downgrading of performance. A novel stepwise method that trains different parts of multi-modal Faster R-CNN using different sub-datasets is presented, showing satisfactory results.
© 2018 The Author(s)
PDF ArticleMore Like This
CNN based OSNR estimation method for long haul optical fiber communication systems
Ziyi Wang, Aiying Yang, Peng Guo, Lihui Feng, and Pinjing He
S4C.3 Asia Communications and Photonics Conference (ACP) 2018
Phase Retrieval Method Based on Deep Learning with Single Image Training in Holographic Data Storage
Jianying Hao, Ruixian Chen, Xiao Lin, Tsutomu Shimura, and Xiaodi Tan
IPDP_01 International Symposium on Imaging, Sensing, and Optical Memory (ODS) 2022
FOG-based Bridge Damage Detection through Deep CNN
Sheng Li, Yan Yang, Lina Yue, Wenbin Hu, Fang Liu, Weibing Gan, and Lixin Wang
Th4.46 Optical Fiber Sensors (OFS) 2020