Table 1
SAD-CNN threshold levels.
| SAD-CNN: Version-1 | |||||
|---|---|---|---|---|---|
| Threshold | Label | Precision | Recall | F1-score | Support |
| Not Det | 0.86 | 0.86 | 0.86 | 50 | |
| 0.5 | Det | 0.87 | 0.87 | 0.87 | 55 |
| average | 0.87 | 0.87 | 0.87 | 105 | |
| Accuracy | 0.87 | 105 | |||
| Not Det | 0.76 | 1.00 | 0.86 | 50 | |
| 0.75 | Det | 1.00 | 0.71 | 0.83 | 55 |
| Average | 0.88 | 0.85 | 0.85 | 105 | |
| Accuracy | 0.85 | 105 | |||
| Not Det | 0.69 | 1.00 | 0.82 | 50 | |
| 0.85 | Det | 1.00 | 0.60 | 0.75 | 55 |
| Average | 0.85 | 0.80 | 0.78 | 105 | |
| Accuracy | 0.79 | 105 | |||
Notes. Summary of the values at different threshold values obtained. The metrics are defined in the Section 3.2. The accuracy of the different threshold levels is put at the bottom of each case. The support represents the quantity of images used to evaluate this metric. The statistic corresponds to the test set of data Group I, with 105 different images edge-on projection at z = 0.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.