Abstract | ||
---|---|---|
•We reformulate 1-bit convolutional nerual networks via a data-driven manner.•A data-adaptive method is proposed to improve 1-bit convolutional neural networks.•A generic module is developed, which can be easily combined with other 1-bit convolutional neural networks.•An efficient binary object detection framework is formulated to balance efficiency and accuracy.•Performance of 1-bit convolutional neural networks on object detection and recognition are enhanced. |
Year | DOI | Venue |
---|---|---|
2022 | 10.1016/j.patrec.2021.12.012 | Pattern Recognition Letters |
Keywords | DocType | Volume |
Deep learning,Model compression,Binary neural networks,Object detection,Object recognition | Journal | 153 |
ISSN | Citations | PageRank |
0167-8655 | 0 | 0.34 |
References | Authors | |
0 | 7 |
Name | Order | Citations | PageRank |
---|---|---|---|
Junhe Zhao | 1 | 3 | 2.07 |
Sheng Xu | 2 | 507 | 71.47 |
Runqi Wang | 3 | 0 | 0.68 |
Baochang Zhang | 4 | 0 | 1.01 |
Guodong Guo | 5 | 2548 | 144.00 |
David Doermann | 6 | 4313 | 312.70 |
Dianmin Sun | 7 | 3 | 3.09 |