Software

Software and Datasets

  • Toolbox Software for single trial detection of Error-related Potentials in brain-machine interfaces: comparison on 11 datasets

Several classification pipelines are systematically compared on 11 datasets to understand whether the classification performance depends more on the classsification pipelines or on data "quality". Waveform morphology and signal-to-noise ratio is also analysed. The software serves also as an implementation guide for BCI researchers.

Available at [https://github.com/mineysm/ErrP_APP]

If you use this software, please refer to:

M. Yasemin, A. Cruz, U. J. Nunes, G. Pires, "Single trial detection of error-related potentials in brain-machine interfaces: A survey and comparison of methods" Journal of Neural Engineering, IOP, december 2022, DOI 10.1088/1741-2552/acabe9 [link - Open Access]


  • Statistical spatial filter 'Fisher Criterion Beamformer' (FCB) (Toolbox: Matlab and Pythons code and application example)

FCB is a statistical spatial filter that allows to increase the discrimination of 2-class event related potentials (ERP) recorded through electroencephalogram (EEG). The Matlab code provides the FCB toolbox and an example to test it on 2 different datasets:

- dataset 1 : P300 oddball elicited in a communication speller (LSC speller)

- dataset 2 : Error-related potentials (ErrP) elicited to detect and correct errors in LSC speller

Available at [FCB_Toolbox (Matlab and Python)]. Made availabe April-June 2022.

Also available on github: https://github.com/gpiresML/FCB-spatial-filter

If you use this toolbox, please refer to:

Gabriel Pires, Urbano Nunes and Miguel Castelo-Branco (2011), "Statistical Spatial Filtering for a P300-based BCI: Tests in able-bodied, and Patients with Cerebral Palsy and Amyotrophic Lateral Sclerosis", Journal of Neuroscience Methods, Elsevier, 2011, 195(2), Feb. 2011: doi:10.1016/j.jneumeth.2010.11.016, https://www.sciencedirect.com/science/article/pii/S0165027010006503?via%3Dihub


  • C-VEP User-identification datasets (ISR-CVEP dataset)

Datasets were recorded using different m-sequences for visual stimulation to validate the concept of User-identification. We provide our own recorded datasets (using a m-sequence of our own and anothed m-sequence tested with 10 participants), and datasets provided elesewhere recorded in the scope of C-VEP BCI that we adapted for C-VEP User-identiification. Available at [C-VEP User-Id dataset: github].

The dataset is associated to paper: F. Roque, G. Pires, J. Perdiz, and U. J. Nunes (2021), A User Identification System based on Code-modulated Visual Evoked Potentials with LED Stimulation, 2021 IEEE International Workshop on Biometrics and Forensics (IWBF), May 2021, Rome, Italy doi: 10.1109/IWBF50991.2021.9465083 [link]


  • Global and Semantic Feature Fusion Approach (GSF2App)

Implementation of the Global and Semantic Feature Fusion Approach (GSF2App) for Indoor Scene Classification using the PyTorch framework. Summary: GSF2App is a two-branch network. In the first branch a state-of-the-art CNN is used to extract global features from the RGB indoor scene image. In the second branch, a semantic feature representation, based on the objects recognized, per frame, by the YOLOv3 approach, were implemented.

Availale at https://github.com/rmca16/GSF2App

If you use this toolbox, please refer to:

R. Pereira, L. Garrote, T. Barros, A. Lopes, U. J. Nunes, A Deep Learning-based Indoor Scene Classification Approach Enhanced with Inter-Object Distance Semantic Features, IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Prague, Czech Republic, Sepetember 2021. [link]


  • EOG image representation and CNN code

A 9-class dataset of Electrooculographic (EOG) signals represented by RGB images and code to train and evaluate a CNN model. Available at [dataset and CNN code: github] .

The dataset is associated to paper: J. Perdiz, L. Garrote, G. Pires and U. J. Nunes (2021), A Reinforcement Learning assisted Eye-driven Computer Game Employing a Decision Tree-based Approach and CNN Classification, IEEE Access, doi: 10.1109/ACCESS.2021.3068055 [link - OpenAccess]


  • DynamicNet Python Class

DynamicNet is a Python class that automatically creates PyTorch Neural Networks (Convolutional-Neural-Network or a Feed-Forward Neural-Network) only specifying in input a dictionary of parameters. Feb 2021. Available at: Dynamic-PyTorch-Net github .

A. Zancanaro - MSc Project "Machine and deep learning algorithms for the analysis of EEG signals for neuroscience and motor rehabilitation", Supervisors: Gabriel Pires, Giulia Cisotto and Urbano Nunes

If you use this toolbox, please refer to:

A. Zancanaro, G. Cisotto, J. R. Paulo, G. Pires, U. J. Nunes, "CNN-based Approaches For Cross-Subject Classification in Motor Imagery: From The State-of-The-Art to DynamicNet", 18th IEEE International Conference on Computational Intelligence in Bioinformatics and Computational Biology 13 to 15 October 2021 - Melbourne, Australia


  • RGB-D Dataset and software

R. Pereira, T. Barros, L. Garrote, A. Lopes, U. J. Nunes, "An Experimental Study of the Accuracy vs Inference Speed of RGB-D Object Recognition in Mobile Robotics", 29th IEEE International Conference on Robot and Human Interactive Communication (RO-MAN), September 2020. [Online]. Available: https://github.com/hcmr-lab/ISR_RGB-D_Dataset


  • Error-related Potentials dataset

Aniana Cruz , Gabriel Pires, Urbano J. Nunes, "Error-related potentials (Primary and secondary ErrP) and P300 event related potentials – BCI-Double-ErrP-Dataset ", IEEE Dataport, March 2020. [Online]. Available: http://dx.doi.org/10.21227/6wpz-g759 . Related to paper published in IEEE TNSRE