Name
Abstract | ||
|---|---|---|
•The harmonic block is designed to learn filter weights in the DCT domain.•Harmonic CNNs are constructed by replacing the convolutional layers.•Parameter learning in frequency domain improves performance.•High-frequency parameter truncation can efficiently compress new or trained CNNs.•The hamonic block can make a CNN invariant to illumination changes. |
Year | DOI | Venue |
|---|---|---|
2022 | 10.1016/j.patcog.2022.108707 | Pattern Recognition |
Keywords | DocType | Volume |
Harmonic network,Convolutional neural network,Discrete cosine transform,Image classification,Object detection,Semantic segmentation | Journal | 129 |
ISSN | Citations | PageRank |
0031-3203 | 0 | 0.34 |
References | Authors | |
4 | 3 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Matej Ulicny | 1 | 1 | 1.03 |
| Vladimir A. Krylov | 2 | 133 | 14.81 |
| Rozenn Dahyot | 3 | 340 | 32.62 |