This is a collaborative development of techniques to classify/individuate macaques based on bioacoustics cues ('coos').
Please check out the Introduction to learn more and get started!
| Epochs | URL | Error rate |
|---|---|---|
| 9 | vanilla DataBlock torchaudio spectrograms | 0.002745 |
| 6 | fastai2 audio xresnet18 | 0.002745 |
| 40 | fastai2 raw audio conv1d | 0.010295 |
| NA | rocket: features from 1d conv + ridge regression | 0.026767 |
| 2 | pretrained resnet18 | 0.063143 |