SkeletonVis : A visualization of RNNs in Skeleton based Action Recognition

The SkeletonVis Tool is a Python software package that is meant as a companion for the paper:

• "Looking Under the Hood: Visualizing What LSTMs Learn" that has been accepted for the 16th International Conference on Image Analysis and Recognition 2019(ICIAR 2019).

The author of this software package is Dhruva Patil , who is currently a Ph.D. student working under the direction of Bruce A. Draper.

The software package has a file called index.py which opens up the URL to be run in the localhost of the web-broswer.

The intention of this software package is for researchers to gain an insight into the working of Long Short Term Memory (LSTM) networks in the domain of Skeleton based Action Recognition.

Use the following link if you wish to download the SkeletonVis software package.

• SkeletonVis Tool

Below is a short video illustrating the use of SkeletonVis to analyze skeletons for a particular case study. Given the video is 1024 pixels across you may with to open it in a new winow.