Use case for the paper submission in IEEE Access: Video sequences of a professional musical ensemble
This repository contains the code that was developed in order to compute Dyadic Synchronization between participants in small groups. We test our system on video sequences of co-performers in a small group of musicians. The src folder contains two files:
- Module_DataExtraction: This file contains the code that extracts the data for each performers by isolating them individually. It stores the output as a .csv file
- Module_DyadicSynchronization: This file contains the code that extracts the data from the .csv file processed by the first file (Module_DataExtraction), and provides the Dyadic Synchronization between all possible pairs in a small group, and in this case the musical ensemble.
Both the files have been shared as a jupyter notebook since this should make it easier to execute the code with more control. The added control helps perform experiments carefully, particularly since the data from pose estimation algorithms can be noisy and sometimes requires manual intervention.
Project is created with:
- Python
- IDE: Jupyter Notebook
- Libraries required: pandas, numpy, matplotlib, scipy, tslearn.metrics, scipy.signal, cmath, math, functools
Our code follows the model as described in the paper, and below is an instance.
As observed, block B (Feature extraction) requires data from a multi-person pose estimation algorithm. For our use case, we utilize AlphaPose v0.4.0. Once the .json file is available as an output, we utilize the file src/Module_DataExtraction.ipynb to create csv files of the data with trajectory information (x,y) of all participants separated in columns.
After this, we utilize the file src/Module_DyadicSynchronization.ipynb to compute the dyadic synchronization between all possible pairs.
We perform a statistical analysis using a two-way ANOVA measure, and utilize Jamovi to conduct these tests.
Sanket Rajeev Sabharwal (sabharwal@edu.unige.it)
If this research helps you please use the following citation:
@ARTICLE{9869836,
author={Sabharwal, Sanket Rajeev and Varlet, Manuel and Breaden, Matthew and Volpe, Gualtiero and Camurri, Antonio and Keller, Peter E.},
journal={IEEE Access},
title={huSync - A Model and System for the Measure of Synchronization in Small Groups: A Case Study on Musical Joint Action},
year={2022},
volume={10},
number={},
pages={92357-92372},
doi={10.1109/ACCESS.2022.3202959}}
S. R. Sabharwal, M. Varlet, M. Breaden, G. Volpe, A. Camurri and P. E. Keller, "huSync - A Model and System for the Measure of Synchronization in Small Groups: A Case Study on Musical Joint Action," in IEEE Access, vol. 10, pp. 92357-92372, 2022, doi: 10.1109/ACCESS.2022.3202959.