ULiège Open Data Repository
Not a member yet
42 research outputs found
Sort by
Dataset & Code for paper: "Multispecies detection and identification of African mammals in aerial imagery using convolutional neural networks"
No full textThis dataset contains aerial images, model result files and the code used in the paper: "Multispecies detection and identification of African mammals in aerial imagery using convolutional neural networks".
It is divided into 3 main parts:
1) The code, which contains the mmdetection v1.0.0 package as well as the adapted version and Jupyter notebooks;
2) The data ('general_dataset.zip'), which contains all the images and annotations of the general dataset used for training the deep learning models in the paper;
3) The results ('results_files.zip'), which contain the models' outputs on the general dataset
Instability strip grid for SPB stars including the effect of the rotation and metallicity
No full textThis dataset contains the results of a grid of stellar evolutionary models computed using the CLES code, along with the corresponding non-adiabatic oscillation computations performed with the MAD non-adiabatic oscillation code. The grid includes the global stellar properties of models with masses ranging from 2.5 to 8.0 solar mass and initial metal mass fractions Z between 0.0125 and 0.025. For each stellar model in the grid, the associated non-adiabatic oscillations have been computed. The dataset provides, for each case, the individual mode frequencies and their corresponding damping or growth rates for example
TopUp Docker
No full textThis a lightweight Docker containerized version of the FSL's "TopUp" tool, cf. https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/topu
Replication Data for: Predictors of completion and outcomes of a 1-month alcohol abstinence campaign in Belgium
No full textThe dataset includes data from 172 participants collected in 2023 and 2024, which are used for statistical analyses in the article entitled "Predictors of completion and outcomes of a 1-month alcohol abstinence campaign in Belgium "
Replication Data for: Noun incorporation in English: A typological perspective
No full textThis dataset comprises Dataset 1 and Dataset 2 that were used in the article “Noun incorporation in English: A typological perspective”, co-authored by Dana Louagie and An Van linden, and published in Functions of Language (2025). Dataset 1 contains 13 comma-separated values (csv) files, one file per N-V lexeme investigated, with each containing 100 items (or less if fewer tokens were found). The data result from a basic sorting of exhaustive samples extracted from all subcorpora of WordBanks Online. The extractions targeted three spelling variants: (i) N and V linked with a hyphen, (ii) N and V in a single graphological word, and (iii) N and V in two graphological words. Dataset 2, in turn, comprises 9 comma-separated values (csv) files containing the results of 9 semi-open searches of N-V combinations in WordBanks Online, with either N or V lexically specified
PeriscOApe ULiège 2018-2020 Data1
No full textThis dataset contains enriched metadata of journal articles published between 2018 and 2020 and for which at least one (co)author is affiliated with the University of Liège.
Data was aggregated and enriched from various sources, incuding OpenAlex and the University of Liège's institutional repository (ORBi). Particular data is then used to create visualizations on the PeriscOApe platform, a website that seeks to monitor the development and various types of Open Acess in the Wallonia-Brussels Federation
High-throughput analysis of Fröhlich-type polaron models
No full textIn this dataset we include data to allow the user to calculate polaron properties, such as zero-point renormalization energy, for a set of materials. There are scripts included that use the standard and the generalized Fröhlich models for the calculation of these properties
EVE: Emotional Validated Expressions, an acted audiovisual corpus
No full textEVE corpus is a resource for speech emotion recognition in English and French audio and audiovisual content. For each language, ten native or near-native actors per language performed 10 linguistically and semantically neutral sentences with different emotions: six basic (fear, anger, happiness, sadness, disgust, surprise), and four complex (confidence, confusion, contempt, empathy). Each was expressed at two arousal levels, with two trials per level. Additionally, each sentence was also produced in a neutral condition, leading to a total of 4,100 recordings. The emotional content of the corpus was perceptually validated by 600 participants per language. The audiovisual recordings were made in high-definition quality following a rigorous methodology, positioning the EVE corpus as a valuable tool for academic and development purposes. Please contact [email protected] . https://camps.aptaracorp.com/ACM_PMS/PMS/ACM/IVA25/49/189787aa-6292-11f0-823f-16ffd757ba29/OUT/iva25-49.html</a
Droplet on a Sugar fiber
No full textThis data set corresponds to the paper submitted to Physical Review Fluids.
The data set include a tab delimited file containing the data from Fig. 3, 4a and 4b. More information will be provided as soon as the paper is accepted
LASLAfiles_Latin_DATformat
No full textWe recommend accessing the Dataverse using the "Tree View". The folder 'LASLA Files DAT format' contains the LASLA Latin Files in a DAT format as described in the documentation found in the subfolder 'DAT_Files_Documentation'. This format is the "historical LASLA format", still used with the LASLA tagger and the tagging interface LEI ("LASLA encoding initiative"). The LASLA files have been converted to the CoNLL-U format and enriched with the links to the LiLa Knowledge Base by the LiLa team: the files are available on Zenodo (https://doi.org/10.5281/zenodo.5961377) and Github (https://github.com/CIRCSE/LASLA).
The files in this repository are shared under the license CC BY-NC-SA 4.0