Implement retraining systems into PyHa
sprestrelski opened this issue · 1 comments
sprestrelski commented
A lot of time during summer has been spent trying to figure out what other people have worked on. Integrating retraining pipelines directly into PyHa would enforce standardization and documentation, making it much more efficient to build off previous work in the project.
Retraining work (* = summer)
- Microfaune (Jacob/Yaman)
- *Microfaune (Sam)
- attempting to update for Tensorflow 2.0 and for PyHa implementation
- TweetyNet (Mugen)
- *TweetyNet (Sean)
- OpenSoundscape (Kushagra)
- OpenSoundscape (Shreyas)
- *OpenSoundscape (Sean)
- replicating Kushagra/Shreya's work, currently retraining on COSMOS dataset
- *BirdNET-Analyzer (Sam)
- added PyHa integration for predictions, working offline on retraining pipeline
sprestrelski commented
Not within scope of PyHa. Most of this is outdated, and any retraining work should go into: https://github.com/UCSD-E4E/acoustic-multiclass-training/