DSpace Collection:http://hdl.handle.net/10174/11352024-03-06T10:16:49Z2024-03-06T10:16:49ZOptimized European Portuguese Speech-To-Text using Deep LearningMedeiros, EduardoCorado, LeonelRato, LuisQuaresma, PauloSalgueiro, Pedrohttp://hdl.handle.net/10174/344662023-02-15T15:56:42Z2022-09-30T23:00:00ZTitle: Optimized European Portuguese Speech-To-Text using Deep Learning
Authors: Medeiros, Eduardo; Corado, Leonel; Rato, Luis; Quaresma, Paulo; Salgueiro, Pedro
Abstract: We have developed an ASR system for European Portuguese implement ing the QuartzNet [3] architecture with the NeMo [4] framework. Two approaches were used in this work: from scratch and using transfer learning. The experiments were data-driven focused instead of algorithm finetuning. Experiments confirm that models developed using transfer learning have shown better results (WER=0.0513) than developing models from scratch (WER=0.1945).2022-09-30T23:00:00ZAutomatic classification of ornamental stones using Machine Learning techniques - a study applied to limestone.Tereso, MarcoRato, LuisGonçalves, Teresahttp://hdl.handle.net/10174/340942023-02-10T11:24:25Z2020-05-31T23:00:00ZTitle: Automatic classification of ornamental stones using Machine Learning techniques - a study applied to limestone.
Authors: Tereso, Marco; Rato, Luis; Gonçalves, Teresa
Abstract: The industry of extraction and transformation of rock minerals has an enormous importance in the Portuguese trade balance. The export volume increases every year, and to maintain these results it is necessary to invest in the modernization and optimization of production processes, as well as, in the classification of raw materials. This study aims to implement a classification model of ornamental rocks through the analysis and classification of images, using machine learning algorithms. The recognition of the type of stone, through the capture of images and subsequent algorithmic analysis, will allow to define quality control scales in future processes, taking into account the different types of stone. In addition, it will also allow to develop models capable of helping in reducing the amount of raw material wasted. This work presents the steps taken to create a classification model, using a dataset of 2260 images distributed over four classes, three of which are very similar to color level and one with a different tone. In this study, the results of the application of three automatic classification algorithms are analyzed. In addition, a discussion of how types of images can improve results and the execution times of algorithms are presented.2020-05-31T23:00:00ZPredicting soil electro-conductivity using Sentinel-1 imagesMedeiros, EduardoGonçalves, TeresaRato, LuisAhmed, Sajibhttp://hdl.handle.net/10174/338872023-02-03T16:03:31Z2021-01-01T00:00:00ZTitle: Predicting soil electro-conductivity using Sentinel-1 images
Authors: Medeiros, Eduardo; Gonçalves, Teresa; Rato, Luis; Ahmed, Sajib
Abstract: The quality and yield of a soil can be measured by using a wide range of
soil indicators. One such indicator is soil’s electro-conductivity which is
an excellent indicator of the presence of soil nutrients. This work aims to
create a machine learning model to predict the soil’s electro-conductivity
(EC) using radar images from the satellite Sentinel-1. Using EC readings
from 14 corn field parcels and Sentinel-1 readings over the course of one
agriculture year, several regression models were generated. These mod-
els were designed using information from the full agriculture year or only
3 months, both or only one of the VV and VH polarisations. The results
show that when using a full year data VV and VH polarisations are able to
generate models with similar performance (R2 of 0.888 for VH and 0.884
for VV) but when using only 3 months data, only April to June trimester
using both polarisations are able to reach similar a performance (R2 of
0.867); moreover VH polarisation seems to carry out more descriptive in-
formation when compared with VV (specially when using only 3 months
Radar data was collected from two time windows each corresponding
data). Finally, performance results seem to be independent of the yearly
radar data time-window.2021-01-01T00:00:00ZSentinel 2 Image Scene Classifica- tion: A Comparison Between Bands and Spectral Indices.Raiyani, KashyapGonçalves, TeresaRato, Luishttp://hdl.handle.net/10174/338832023-02-03T16:00:14Z2021-01-01T00:00:00ZTitle: Sentinel 2 Image Scene Classifica- tion: A Comparison Between Bands and Spectral Indices.
Authors: Raiyani, Kashyap; Gonçalves, Teresa; Rato, Luis
Abstract: Given the continuous increase in the global population, the food manufacturers are advocated to either intensify the use of cropland or expand the farmland, making land cover and land usage dynamics mapping vital
in the area of remote sensing. In this regard, identifying and classifying a high-resolution satellite imagery scene is a prime challenge. Several approaches have been proposed either by using static rule-based thresholds (with limitation of diversity) or neural network (with data-dependent limitations). This paper adopts an inductive approach to build classifiers from spectral reflectances, comparing usefulness of the various spectral indices to raw bands information. More specifically, it considers Sentinel2 data for six classes Scene Classification (Water, Shadow, Cirrus, Cloud, Snow and Other). The experimental results show that using raw bands
performs equally well, claiming that raw bands information can be used as a replacement of the spectral indices.2021-01-01T00:00:00Z