Research Article
Concrete Surface Crack Recognition Based on Coordinate Attention Neural Networks
Table 1
Experimental parameters of CaNet.
| Parameter type | Parameter name | Parameter setting |
| Hardware environment | Operating system | Windows 10 (64 bit) | Version number | 20H2 | RAM | 64 GB(63.8 available) | Processor | Intel(R) core(TM) i7-9700u | Graphics card | 2080ti GPU |
| Software environment | CaNet | Python 3.8.12 | Pytorch 1.8.2 | CUDA 10.2 |
| Training parameter setting | Batch size(training) | 32 | Batch size(testing) | 8 | Learning rate | | Epoches | 140 | Optimizer | Adam optimizer | Loss function | Cross entropy loss | Training method | Incremental gradient descent |
|
|