Table 1.
Method | F(BSDS300) | AP(BSDS300) | F(BSDS500) | AP(BSDS300) | ||
---|---|---|---|---|---|---|
Human | 0.79 | - | 0.80 | - | ||
Machine learning | Shallow | Pb | 0.63 | - | 0.67 | - |
BEL | 0.65 | - | 0.61 | - | ||
Deep | gPb | 0.70 | 0.66 | 0.71 | 0.65 | |
Deep Edge | - | - | 0.75 | 0.80 | ||
HED | - | - | 0.78 | 0.83 | ||
Low-level features | Classical edge detection | Canny | 0.58 | 0.58 | 0.61 | 0.58 |
Normalised Cuts | 0.62 | 0.42 | 0.63 | 0.45 | ||
Biological | SCO | 0.66 | 0.70 | 0.67 | 0.71 | |
MCI | 0.62 | - | 0.64 | - | ||
Ours | 0.68 | 0.70 | 0.70 | 0.74 |