Table 3 Architecture of the proposed discriminator.
From: CalliFormer: a structure-aware transformer for Chinese calligraphy generation
Layer | Type | Shape |
|---|---|---|
Input | Image pair | 2 × 256 × 256 |
L1 | Conv5 × 5, BN, LeakyReLU | 64 × 128 × 128 |
L2 | Conv5 × 5, BN, LeakyReLU | 128 × 64 × 64 |
L3 | Conv5 × 5, BN, LeakyReLU | 256 × 32 × 32 |
L4 | Conv5 × 5, BN, LeakyReLU | 512 × 16 × 16 |
L5 | Linear, Sigmoid | 1 |
