Titulo Estágio
Automatic Gesture Recognition Tools for Breaking (aka Breakdance)
Áreas de especialidade
Sistemas de Informação
Sistemas Inteligentes
Local do Estágio
DEI
Enquadramento
Breaking (commonly known as Breakdance) is an acrobatic street dance that involves a highly dynamic and complex set of movements. It poses significant challenges for effective movement recognition and analysis. With its inclusion in the 2024 Paris Olympics, Breaking will require more structured and thoughtful training to improve performance execution. The competition sets of movements/exercises must be studied to understand gaps and where to improve. However and to this date, there have been no studies or technological support for Breaking movement recognition or analysis. This dissertation aims to explore tools for automatic gesture recognition based on Breaking videos to develop an application (model and tools) that identifies and classifies Breaking movements. This work aims to contribute to the sports technology and movement recognition areas, specifically addressing Breaking technology and context.
Objetivo
The goal of this dissertation is to investigate existing motion analysis methodologies applied to sports and dance to develop an application (model and tools) that identify Breaking movements based on videos uploaded by Bboys/Bgirls. To achieve this, a preliminary study will be conducted to identify the existing gesture recognition algorithms. The most promising options will then be tested, leading to the creation of an application to analyse uploaded Breaking videos, identify the movements performed by Bboys/Bgirls and produce a movement database.
Plano de Trabalhos - Semestre 1
1 - State of the art review of models and tool for sports and dance video analysis
2- Gain insight into the different types of Breaking movements and the main challenges of acrobatic and intrinsic movement recognition/analysis
3 - Review of existent gesture recognition algorithms
4 - Collection of Breaking videos to train and test the selected approaches/methodologies
5 - Writing intermediate document of the dissertation.
Plano de Trabalhos - Semestre 2
6 - Iterative development and evaluation of the gesture recognition algorithms
7 - Testing, experimentation and validation of the gesture recognition algorithms with breaking videos datasets.
8 - Analysis and reporting of the results.
9 - Dissertation writing
A - Writing and submission of a scientific article (Optional and contingent on the significance of the results)
Condições
The student is expected to have an interest in Sports Technology and any prior experience related to Breaking or similiar sports/dance is a plus. Previous knowledge of machine learning tools is also welcome.
Observações
This work will be linked to the PhD project of Leonor Portugal da Fonseca, who is “Exploring how technology can support Breaking training”.
The student will have access to professional Bboys/Bgirls if desired, and any questions regarding Breaking can be discussed with the team.
The work will be co-supervised by Leonor Portugal da Fonseca and Paulo Peixoto from DEEC-UC
Orientador
Paula Silva
paulasilva@dei.uc.pt 📩