Tactile perception of the material properties in real-time using tiny embedded systems is a challenging task and of grave importance for dexterous object manipulation such as robotics, prosthetics and augmented reality [1-4] . As the psychophysical dimensions of the material properties cover a wide range of percepts, embedded tactile perception systems require efficient signal feature extraction and classification techniques to process signals collected by tactile sensors in real-time.
There are four CSV files (X, Y, Z, and S) in the dataset corresponding to the sensor recordings. The 3-dimensional accelerometer sensor recordings are denoted by X, Y, and Z, respectively. The sound recordings from the electret condenser microphone are denoted by S. As there are 12 classes in the dataset, there is one line for each class in the CSV files. For each texture, 20 seconds of recordings are collected. Therefore, each line in the X, Y, Z files has 4,000 samples (20 sec x 200Hz sampling rate) and each line in the S file has 160,000 samples (20 sec x 8 kHz). The training and test sets for the machine learning classifiers can be created by snipping short frames out of these recordings and applying signal feature extraction. For example, the first 400 columns of the 12th row of X.csv and the first 16,000 columns of the 12th row of S.csv both correspond to the first 2 seconds of the recordings for texture class 12. The Python programs we have developed will be made available upon request.
Please cite the dataset and accompanying paper if you use this dataset:
- Kursun, O. and Patooghy, A. (2020) "An Embedded System for Collection and Real-time Classification of a Tactile Dataset", IEEE Access (accepted for publication).
- Kursun, O. and Patooghy, A. (2020) "Texture Dataset Collected by Tactile Sensors", IEEE Dataport, 2020.
The training, validation, and test set used for Deep Xi (https://github.com/anicolson/DeepXi).
This is the noisy-speech test set used in the original Deep Xi paper: https://doi.org/10.1016/j.specom.2019.06.002. The clean speech and noise used to create the noisy-speech set are also available.
The dataset consists of two populations of fetuses: 160 healthy and 102 Late Intra Uterine Growth Restricted (IUGR). Late IUGR is an adverse pathological condition encompassing chronic hypoxia as a consequence of placental insufficiency, resulting in an abnormal rate of fetal growth. In standard clinical practice, Late IUGR diagnosis can only be suspected in the third trimester and ultimately confirmed at birth. This data collection comprises of a set of 31 Fetal Heart Rate (FHR) indices computed at different time scales and domains accompanied by the clinical diagnosis.
The data for healthy and Late IUGR populations are included in a single .xlsx file.
Participants are listed by rows and features by columns. In the following we report an exhaustive list of features contained in the dataset accompanied by their units, time interval employed for the computation, and scientific literature references:
Fetal and Maternal Domains
- Clinical Diagnosis [HEALTHY/LATE IUGR]: binary variable to report the clinical diagnosis of the participant
- Gestational Age [days]: gestational age at the time of CTG examination
- Maternal Age [years]: maternal age at the time of CTG examination
- Sex [Male (1)/Female (2)]: fetal sex
Morphological and Time Domains
- Mean FHR [bpm] – 1-min epoch: the mean of FHR excluding accelerations and decelerations
- Std FHR [bpm] – 1-min epoch: the standard deviation of FHR excluding accelerations and decelerations
- DELTA [ms] – 1-min epoch: defined in accordance with ,  excluding accelerations and decelerations
- II  – 1-min epoch: defined in accordance with ,  excluding accelerations and decelerations
- STV [ms] – 1-min epoch: defined in accordance with ,  excluding accelerations and decelerations
- LTI [ms] – 3-min epoch: defined in accordance with ,  excluding accelerations and decelerations
- ACC_L [#] – entire recording: the count of large accelerations defined in accordance with , 
- ACC_S [#] – entire recording: the count of small accelerations defined in accordance with , 
- CONTR [#]– entire recording: the count of contractions defined in accordance with , 
- LF [ms²/Hz] – 3-min epoch: defined in accordance with , LF band is defined in the range [0.03 - 0.15] Hz
- MF [ms²/Hz] – 3-min epoch: defined in accordance with , MF band is defined in the range [0.15 - 0.5] Hz
- HF [ms²/Hz] – 3-min epoch: defined in accordance with , HF band is defined in the range HF [0.5 - 1 Hz]
- ApEn [bits] – 3-min epoch: defined in accordance with , m = 1, r = 0.1*standard deviation of the considered epoch
- SampEn [bits] – 3-min epoch: defined in accordance with , m = 1, r = 0.1*standard deviation of the considered epoch
- LCZ_BIN_0 [bits] – 3-min epoch: defined in accordance with , binary coding and p = 0
- LCZ_TER_0 [bits] – 3-min epoch: defined in accordance with , tertiary coding and p = 0
- AC/DC/DR [bpm] – entire recording: defined in accordance with –, considering different combinations of parameters T and s, L is constant and equal 100 samples; e.g, AC_T1_s2 is defined as the acceleration capacity computed setting the parameters T = 1 and s = 2
 D. Arduini, G. Rizzo, A. Piana, P. Bonalumi, P. Brambilla, and C. Romanini, “Computerized analysis of fetal heart rate—Part I: description of the sys- tem (2CTG),” J Matern Fetal Invest, vol. 3, pp. 159–164, 1993.
 M. G. Signorini, G. Magenes, S. Cerutti, and D. Arduini, “Linear and nonlinear parameters for the analysis of fetal heart rate signal from cardiotocographic recordings,” IEEE Trans. Biomed. Eng., vol. 50, no. 3, pp. 365–374, 2003.
 FIGO, “Guidelines for the Use of Fetal Monitoring,” Int. J. Gynecol. Obstet., vol. 25, pp. 159–167, 1986.
 R. Rabinowitz, E. Persitz, and E. Sadovsky, “The relation between fetal heart rate accelerations and fetal movements.,” Obstet. Gynecol., vol. 61, no. 1, pp. 16–18, 1983.
 S. M. Pincus and R. R. Viscarello, “Approximate entropy: a regularity measure for fetal heart rate analysis.,” Obstet. Gynecol., vol. 79, no. 2, pp. 249–55, 1992.
 D. E. Lake, J. S. Richman, M. P. Griffin, and J. R. Moorman, “Sample entropy analysis of neonatal heart rate variability,” Am. J. Physiol. - Regul. Integr. Comp. Physiol., vol. 283, no. 3, pp. R789–R797, 2002.
 A. Lempel and J. Ziv, “On the complexity of finite sequences,” IEEE Trans. Inf. Theory, vol. 22, no. 1, pp. 75–81, 1976.
 A. Bauer et al., “Phase-rectified signal averaging detects quasi-periodicities in non-stationary data,” Phys. A Stat. Mech. its Appl., vol. 364, pp. 423–434, 2006.
 A. Fanelli, G. Magenes, M. Campanile, and M. G. Signorini, “Quantitative assessment of fetal well-being through ctg recordings: A new parameter based on phase-rectified signal average,” IEEE J. Biomed. Heal. Informatics, vol. 17, no. 5, pp. 959–966, 2013.
 M. W. Rivolta, T. Stampalija, M. G. Frasch, and R. Sassi, “Theoretical Value of Deceleration Capacity Points to Deceleration Reserve of Fetal Heart Rate,” IEEE Trans. Biomed. Eng., pp. 1–10, 2019.
The present database contains records of underwater sounds produced by dolphins of the species Tursiops Truncatos. The dolphins live in Dolphinarium Varna (Varna, Bulgaria) and are a family of five individuals. The records was made in the autumn 2019 by the “SigNautic Lab” crew, using measurement equipment of the world’s leading suppliers Bruel&Kjaer and National Instruments. The database contains two sets of .wav audio lossless uncompressed files – 13 “bursts” and 104 “clicks” with durations of 250 ms to 9 s. All signals are sampled at 250 kHz and are amplitude-normalized.
Please, read the corresponding Matlab software\Read me.txt and Data Acquisition and Signal Analysis Parameters.pdf files.
BCI-Double-ErrP-Dataset is an EEG dataset recorded while participants used a P300-based BCI speller. This speller uses a P300 post-detection based on Error-related potentials (ErrPs) to detect and correct errors (i.e. when the detected symbol does not match the user’s intention). After the P300 detection, an automatic correction is made when an ErrP is detected (this is called a “Primary ErrP”). The correction proposed by the system is also evaluated, eventually eliciting a “Secondary ErrP” if the correction is wrong.
A detailed description of the data is given in “BCI-Double-ErrP-Dataset-instructions.pdf” and a Matlab code example is provided to extract P300 and ErrPs (primary and secondary).
There are 4 folders, one with the datasets of the P300 calibration (session 1), one with the datasets of the ErrP calibration (session 1), one with the datasets of the testing session (session 2), and a folder with the Matlab code to run the example.
In the present article we analyze data from two temperature sensors of the Curiosity rover, which has been active in Mars since August 2012. Temperature measurements received from the rover are noisy and must be processed and validated before being delivered to the scientific community. Currently, a simple moving average filter is used to perform signal denoising. The application of this basic algorithm is based on the assumption that the noise is stationary and statistically independent from the underlying structure of the signal, an arguable assumption in this kind of harsh environment.
CUPSNBOTTLES is an object data set, recorded by a mobile service robot. There are 10 object classes, each with a varying number of samples. Additionally, there is a clutter class, containing samples where the object detector failed.
Download and extract the ZIP file containing all files. There is python code available (under 'scripts') to easily load the data set. Other programming languages should also handle .jpg, .hdf and .csv files for easy access. For easy access with python, a pickle dump file has been added. This has no extra information compared to the .csv file.