Research Article
Handwritten Geez Digit Recognition Using Deep Learning
Table 3
The CNN models for Geez handwritten digit recognition.
| | Model | Layer 1 | Layer 2 | Layer 3 | Layer 4 | Layer 5 | Layer 6 | Layer 7 | Layer 8 | Layer 9 | Layer 10 | Layer 11 | Layer 12 |
| | 1 | Conv 3 × 3@1@32 and ReLU | Conv 3 × 3@1@32 and ReLU | Max-pooling 2 × 2@2@32 | Conv 3 × 3@1@64 and ReLU | Conv 3 × 3@1@64 and ReLU | Max-pooling 2 × 2@2@64 and dropout | Conv 3 × 3@1@64 and ReLU | Conv 3 × 3@1@128 and ReLU | Max-pooling 2 × 2@2@128 and dropout | | | | | 2 | Conv 3 × 3@1@32 and ReLU | Conv 3 × 3@1@32 and ReLU | Max-pooling 2 × 2@2@32 | Conv 3 × 3@1@32 and ReLU | Conv 3 × 3@1@32 and ReLU | Max-pooling 2 × 2@2@32 | Conv 3 × 3@1@64 and ReLU | Conv 3 × 3@1@64 and ReLU | Max-pooling 2 × 2@2@64 and dropout | Conv 3 × 3@1@64 and ReLU | Conv 3 × 3@1@64 and ReLU | 3 additional layers such as pooling, conv, pooling | | 3 | Conv 3 × 3@1@32 and ReLU | Conv 3 × 3@1@32 and ReLU | Max-pooling 2 × 2@2@32 | Conv 3 × 3@1@32 and ReLU | Conv 3 × 3@1@32 and ReLU | Max-pooling 2 × 2@2@32 and dropout | Conv 3 × 3@1@64 and ReLU | Conv 3 × 3@1@64 and ReLU | Conv 3 × 3@1@64 and ReLU | Max-pooling 2 × 2@2@64 | Conv 3 × 3@1@64 and ReLU | Convolution layer with pooling and dropout layer | | 4 | Conv 3 × 3@1@32 and ReLU | Conv 3 × 3@1@32 and ReLU | Conv 3 × 3@1@32 and ReLU | Max-pooling 2 × 2@2@32 | Conv 3 × 3@1@64 and ReLU | Conv 3 × 3@1@64 and ReLU | Conv 3 × 3@1@64 and ReLU | Max-pooling 2 × 2@2@64 and dropout | Conv 3 × 3@1@128 and ReLU | Max-pooling 2 × 2@2@128 and dropout | | | | 5 | Conv 3 × 3@1@32 and ReLU | Conv 3 × 3@1@32 and ReLU | Conv 3 × 3@1@64 and ReLU | Max-pooling 2 × 2@2@64 and dropout | Conv 3 × 3@1@64 and ReLU | Conv 3 × 3@1@64 and ReLU | Conv 3 × 3@1@128 and ReLU | Max-pooling 2 × 2@2@128 and dropout | | | | | | 6 | Conv 3 × 3@1@32 and ReLU | Conv 3 × 3@1@32 and ReLU | Conv 3 × 3@1@32 and ReLU | Max-pooling 2 × 2@2@32 | Conv 3 × 3@1@64 and ReLU | Conv 3 × 3@1@64 and ReLU | Conv 3 × 3@1@64 and ReLU | Max-pooling 2 × 2@2@64 and dropout | Conv 3 × 3@1@64 and ReLU | Max-pooling 2 × 2@2@64 and dropout | Conv 3 × 3@1@128 and ReLU | Max-pooling 2 × 2@2@128 and dropout |
|
|
