Table 5 Quantitative detection performance for two EM datasets
From: Effective automated pipeline for 3D reconstruction of synapses based on deep learning
Metrics | Sun [18] | U-Net [23] | F-RCNN | F-RCNN + | F-RCNN + | F-RCNN + fusion + |
| Â | Â | Â | Â | fusion | continuity | continuity |
ATUM-SEM DATASET | Â | Â | Â | |||
AP | 67.6% | 74.9% | 84.1% | 89.7% | 86.8% | 92.8% |
F1 | 64.8% | 74.8% | 79.3% | 83.6% | 83.9% | 89.2% |
FIB-SEM DATASET | Â | Â | Â | |||
Metrics | Becker [15] | U-Net [23] | F-RCNN | F-RCNN + | F-RCNN + | F-RCNN + fusion + |
| Â | Â | Â | Â | fusion | continuity | continuity |
AP | 90.6% | 82.4% | 90.9% | 91.9% | 92.4% | 93.5% |
F1 | 88.4% | 82.0% | 88.5% | 90.6% | 89.2% | 91.4% |