This repository contains code to generalize autoencoder based analysis of behavioral video data, based on work previously described in Wang et al. 2023. This repository will be used to develop a standalone version of an autoencoder which would be more easily applicable to other datasets.
- Standalone backbone of autoencoder model.
- Different architecture variants as may be useful for multi-session data recordings
- Data loading modules
- Frame based data loading
- Tar based data loading
- Video based data loading
- Preprocessing workflows
- Uniform subsetted frame selection
- Frame selection with trialwise sequences.