- Added torch.PyTorch: Apply a PyTorch module. (#95, #106)
- Added batch.BatchedPipeline: Combine consecutive objects into a batch. (#92, #106)
- Added mjpeg_streamer.MJPEGStreamer: Stream images via HTTP (e.g. to the Browser). (#75)
- Added integration.raspi.PiCameraReader: Read frames from the Raspberry Pi's camera. (#75)
- Added filters as a replacement for stat. (#77)
- Added utils.StreamEstimator to estimate the remaining length of a stream. (#79, #91)
- Added utils.stream_groupby: Split a stream into sub-streams by key.
- Added support for Python 3.9 and 3.10 (#87).
- Added HDF5Writer. (#117)
- Added scalebar: Append scalebars to images. (#113)
- ecotaxa: Return EcotaxaObject in EcotaxaReader (#102). Allow Variables for archive_fn in EcoTaxaWriter (#100). More improvements (#109).
- Use UnavailableObject instead of import_optional_dependency.
- Make pandas and tqdm a hard dependency.
- Require scikit-image>=0.19 (#86)
- Deprecate stat (#77).
- Drop import_optional_dependency copied over from pandas.
- Dropped support for Python 3.6, 3.7 and 3.8 (#87, #119).
- calculation of n_remaining_hint in stream.Slice (#111).
- ValueError: 'version' argument is required in Sphinx directives (#80).
- UnknownArchiveError: Close EcoTaxa archives (#88).
- wrongly reported n_remaining_hint in Progress after Slice (#105).