Releases: SchwarzNeuroconLab/DeepLabStream
Advanced behavioral classification
- full integration of online classification of user-defined behavior using SiMBA and B-SOID.
- SOCIAL CLASSIFICATION with SiMBA 14bp two animal classification (more to come!)
- Unsupervised Classification with B-SOID
- New wiki guide and example experiment to get started with online classification: Advanced Behavior Classification
- this version has new requirements (numba, pure, scikit-learn), so be sure to install them (e.g.
pip install -r requirements.txt
).
Communication Biology Version
v1.3.1-archive hotfix for dlc issue with pose/peaks
Additional Networks for real-time pose estimation
This release integrated additional pose estimation networks into DlStream.
Networks include DeepLabCut based pose estimation (DLC-Live, maDLC) and DeepPoseKit based pose estimation (StackedDenseNet, StackedHourGlass and Leap). They can be fully utilized in DLStream experiments (with some limitations to maDLC), but have not been fully benchmarked. This is a beta release. We will be continuing to implement other pose estimation networks (feedback is welcome) and evaluate the performance of each type of network (contributions welcome) as we go to the full release. We updated the settings, so be sure to check out the updated Installing&Testing section.
So stay tuned and try out the new features!
Minor Update: Settings
In this update we simplified the settings file:
STREAMING_SOURCE now directly selects the input source (camera, video, ipwebcam)
Also:
Preparations for future releases! Stay tuned!
Minor Updates and Features
Release of additional GPIO stimulation control modules:
-
Raspberry Pi support
-
Arduino Support
Read more in the wiki.
Major Release: Out-of-the-Box Experiment design
This release includes the following features:
design_experiment and config.ini support for experiments from the base modules
updated custom experiment call (with no need to change the core code)
several new adaptive experiments, triggers and stimulations available
ip webcam support
prerecorded video support
minor bug fixes
Minor Updates and Features
We included a VideoManager that allows the integration of prerecorded videos into the DLStream app. Previously this was only possible in Videoanalyzer.py.
This included new settings:
Video (TRUE/FALSE)
minor update:
when using --benchmark-enabled, the output txt file is now automatically named after the entries in the settings
Full experiment and trigger release
In this release we added our full experimental library, updated the readme and wiki to incorporate further guides and tutorials.
BioRxiv Version
DLstream Version associated with the preprint on BioRxiv.