Calisthenics Skills Temporal Video Segmentation

This repository hosts the keypoints dataset and some additional codes related to the paper:
Finocchiaro, A.; Farinella, G. and Furnari, A. (2024). Calisthenics Skills Temporal Video Segmentation. In Proceedings of the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 2: VISAPP, ISBN 978-989-758-679-8, ISSN 2184-4321, pages 182-190.

Other works involved:

[JVCI 2018] A. Furnari, S. Battiato, G. M. Farinella, Personal-Location-Based Temporal Segmentation of Egocentric Video for Lifelogging Applications . Journal of Visual Communication and Image Representation , 52 , pp. 1-12

Repository Overview

Data

The /data directory you can find the keypoints_dataset, presented and discussed in the paper.

Source Code

The /src directory contains the source code organized into three subfolders:

• inference_scripts

• codec.py: This script encodes and decodes labels, essential for the inference process.

• inference.py: The main inference script. It tests the entire pipeline, taking a video's relative path as an argument and displaying corresponding segments with their times.

• openpose_script.py: This script extracts keypoints from a video and is utilized within the inference script.

• model

• mlp.py: This script defines the architecture of the multilayer perceptron (MLP), encompassing both training and testing phases.

• temporal_segmentation

• furnari2018.py: Implementation of the Probabilistic algorithm.

• heuristic.py: Implementation of the Heuristic algorithm.

Requirements

To execute inference on a new video, begin by installing the required libraries listed in the requirements.txt file, running the following command.

pip3 install -r requirements.txt

Then, you will need to install OpenPose in your computer, all the steps for its installation are listed in the following link:

OpenPose installation guide