Software Used in HUGS
Overview
Navigate the subpages for an inventory of the software we use in HUGS and the capacity in which we use it.
Software and instructions for decomposing 3D video files generated by the Intel RealSense camera
Free and custom software
May be needed if videos beyond those in the Infants folder are needed for additional exploration
To develop HUGS-3 app
Free
Used to direct outputs from HUGS-3 to Android over Bluetooth
Free
Labeling of selected clips to train algorithm (Supervised Learning)
Free
Possibly a bit complicated to install.
Allows cloud processing of training/analysis in DLC, analysis in OpenPose.
"On-device machine learning for everyone"
OpenPose
Deep learning algorithm trained on adult skeleton model, including hands
Free
Run the "demo" program in Colab with lines of code modified to substitute videos we are interested in
Output needs to be processed in R to compare with DeepLabCut output
Allows browsing of .AVI files to identify postures of interest for DLC
Greater precision than the DLC GUI provides
Very useful tool – we don’t expect mastery, just ability to leverage for our specific tasks
Allows extraction of labeled files to match frames labels vs frames processed
R, OpenPoseR, & PURRR
Statistical software with amazing file processing capabilities
Needed, along with OpenPoseR and PURRR packages, to rewrite JSON files to .csv files for processing to compare with DLC
Free, online, interactive video annotation tool
Potential alternative to the DeepLabCut GUI for labeling training sets
Free, HUGS-LAB has shared repository online
Zotero is citation software. Learning Zotero will greatly streamline your reference building and manipulation and increase team efficiency
Summer 2023 Group: SmartSuite MVP Development