Visual Dataset of Traditional Portuguese Musical Instruments
Description
This dataset was created to support classifying and preserving traditional Portuguese musical instruments using computer vision techniques. The hypothesis is that visual data when systematically organized and labelled, can effectively train machine learning models to recognize and classify Portuguese folk instruments. The dataset includes frames extracted from videos by 'A Música Portuguesa A Gostar Dela Própria' (MPAGDP), an association focused on documenting Portuguese cultural heritage. Five frames per video were selected and normalized to ensure consistency, enabling accurate analysis and model training. The data can be used to identify features of Portuguese instruments, supporting applications in digital ethnomusicology, automated classification, and cultural studies. Researchers can interpret the data as a tool for deep learning and comparative analysis of visual characteristics across instrument types. This dataset provides a structured foundation for studies in cultural preservation, deep learning, and visual pattern recognition.
Files
Steps to reproduce
The data were obtained by extracting frames from videos provided by 'A Música Portuguesa A Gostar Dela Própria' (MPAGDP), a cultural association dedicated to documenting Portuguese traditional music. Five representative frames were selected for each video featuring a Portuguese instrument to capture visual details. Frames were extracted and normalized to ensure consistent across the dataset. Each image was labelled using the Roboflow tool, creating a structured dataset suitable for training and testing in computer vision applications. This workflow can be reproduced by following similar extraction, selection, and labelling processes on publicly available video content documenting traditional instruments.