Table 3 Comparing the models with and without the output branching design.
From: Learning from algorithm-generated pseudo-annotations for detecting ants in videos
Model | HA (%) | AA (%) | Precision | Recall | \(F_1\) score |
|---|---|---|---|---|---|
Using Full HA | 100 | 0 | 0.8760 | 0.7520 | 0.8092 |
w/o output branching | 2 | 98 | 0.8383 | 0.7006 | 0.7633 |
With output branching | 2 | 98 | 0.8482 | 0.7142 | 0.7754 |
w/o output branching | 10 | 90 | 0.8321 | 0.7493 | 0.7885 |
With output branching | 10 | 90 | 0.8631 | 0.7621 | 0.8090 |
