Open bgiebl opened 1 year ago
The gradient magnitude map is defined as $g(\mathbf{I}) = \sqrt{(\mathbf{I} \mathbf{h}_x) + (\mathbf{I} \mathbf{h}_y)}$
Why are you using g_I = edge_filter(median_filter(x)) in lines 35 and 36 in gms_loss.py instead?
g_I = edge_filter(median_filter(x))
The gradient magnitude map is defined as $g(\mathbf{I}) = \sqrt{(\mathbf{I} \mathbf{h}_x) + (\mathbf{I} \mathbf{h}_y)}$
Why are you using
g_I = edge_filter(median_filter(x))in lines 35 and 36 in gms_loss.py instead?