Table 2 Precision and recall values for USV segmentation obtained on the test set of our dataset, considering \(t_{\text {IoU}} = 0.6\). Comparison between AE, UNET, RNN, A-MUD, DS, DS with de-noiser, USVSEG and HM. \(t_c = 0.5\) for our methods.
From: Extended performance analysis of deep-learning algorithms for mice vocalization segmentation
| Â | AE | UNET | RNN | A-MUD | DS | DS (w/ de-noiser) | USVSEG | HM |
|---|---|---|---|---|---|---|---|---|
Precision | 90.14% | 91.11% | 86.18% | 90.62% | 66.37% | 75.75% | 85.72% | 82.03% |
Recall | 90.80% | 92.08% | 84.15% | 80.03% | 63.72% | 63.66% | 87.99 % | 71.18% |
