1. Repositório Aberto da Universidade do Porto
2. FEUP - Faculdade de Engenharia
3. FEUP - Dissertação

Author(s):	Silvia Helena Lemba Tavares
Title:	Hand Gesture Detection and Classification in Sign Language Video Using Skeleton-Based Action Recognition
Issue Date:	2025-10-14
Description:	This dissertation aims to detect hand gestures in sign language videos using skeleton-based models where each video represents a word in sign language. The goal is to focus on the body's structure and movement, providing a more accurate solution in hand pose detection. For the initial phase, the state of the art is done to identify and summarize the latest research in this field. Subsequently, the following steps are to be applied: Video processing involves extracting frames from videos and using libraries such as OpenPose or Mediapipe to detect and extract the positions of hand joints. Data preprocessing includes normalizing joint positions and organizing them into temporal sequences that represent signs. Model training utilizes convolutional neural networks (CNNs) and recurrent neural networks (RNNs), such as LSTMs or GRUs, for gesture recognition, with the models being trained using TensorFlow or PyTorch. Finally, implementation and real-time use focus on optimizing the model for real-time performance and integrating it into an application.
Subject:	Outras ciências da engenharia e tecnologias Other engineering and technologies
Scientific areas:	Ciências da engenharia e tecnologias::Outras ciências da engenharia e tecnologias Engineering and technology::Other engineering and technologies
TID identifier:	204119537
URI:	https://hdl.handle.net/10216/170484
Document Type:	Dissertação
Rights:	openAccess
Appears in Collections:	FEUP - Dissertação

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.