Table 1.

Speeds of traditional method and neural network method. RIE has a window size of 4 $\times$ 4 and a quantization level of 8. The subscript $\mathrm{t}$ denotes the traditional method, and $\mathrm{nn}$ denotes the method used in this paper.

Metrics	$\mathbf{G}\mathbf{I}{\mathbf{E}}_{\mathbf{t}}$	$\mathbf{G}\mathbf{I}{\mathbf{E}}_{\mathbf{n}\mathbf{n}}$	$\mathbf{R}\mathbf{I}{\mathbf{E}}_{\mathbf{t}}$	$\mathbf{R}\mathbf{I}{\mathbf{E}}_{\mathbf{n}\mathbf{n}}$
Time/ms	56.2	14.5	138,600	25.9