Table 6.
Evaluated model | Training set size (No. of images) | Test set size (No. of images) | Fine‐grained classification accuracy | Coarse classification accuracy | Accuracy when tolerating a [±1] error range | Accuracy when tolerating a [±2] error range | Fine‐grained mean L1 error | Coarse mean L1 error |
---|---|---|---|---|---|---|---|---|
ResNet50‐Pheno | 16,298 | 4073 | 43.4% | 69.0% | 67.1% | 82.8% | 1.35 | 0.37 |