This is the example dataset and source code of the paper "A Generalized Channel Dataset Generator for 5G New Radio Systems Based on Ray-Tracing" , and the data generator can be downloaded for free by researchers.

The datagenerator_raytrace3.0 is now support for mm-wave (28GHz+) channel state information.

The advantages of our dataset are as follows:

  • To deal with the large-scale dataset requirements, this letter models the channel according to 5G NR stan- dard published by 3GPP, and provides a 5G NR channel datasets generator based on CDL channel model.


Opportunity++ is a precisely annotated dataset designed to support AI and machine learning research focused on the multimodal perception and learning of human activities (e.g. short actions, gestures, modes of locomotion, higher-level behavior).


Complete documentation is provided in the readme.


Detecting radioactive materials in mixtures is challenging due to low concentration, environmental factors, sensor noise, and others. This paper presents new results on nuclear material identification and mixing ratio estimation for mixtures of materials in which there are multiple isotopes present. Conventional and deep learning-based machine learning algorithms were compared. Both simulated and actual experimental data were used in the comparative studies.


This is a data set about seven letter gestures


This is a data set of delay timer design for paper ''Design of delay timers based on estimated probability mass functions of alarm durations''.


The SiCWell Dataset contains data of battery electric vehicle lithium-ion batteries for modeling and diagnosis purposes. In this experiment, automotive-grade lithium-ion pouch bag cells are cycled with current profiles plausible for electric vehicles. 

The analysis of current ripples in electric vehicles and the corresponding aging experiments of the battery cells result in a dataset, which is composed of the following parts: 



Cell Aging Scenarios

The battery cells are cycled in groups of three cells in series. The scenarios for each cell are the following:

  • Ka01, Ka02: Calendar test 35°C 80% SoC
  • Ka03, Ka04: Calendar test 35°C 45% SoC
  • Ka05, Ka06: Calendar test 45°C 80% SoC
  • Ka07, Ka08: Calendar test 45°C 20% SoC
  • Ka09, Ka10: Calendar test 45°C 45% SoC
  • Ka11, Ka12: Calendar test 45°C 60% SoC
  • DC01, DC02, DC03: DC cycling
  • AC01, AC02, AC03: Sinusoidal cycling 10 kHz, 12.5 A
  • AC04, AC05, AC06: Sinusoidal cycling 10 kHz, 25.0 A
  • AC07, AC08, AC09: Sinusoidal cycling 10 kHz, 6.25 A
  • AC10, AC11, AC12: Sinusoidal cycling 40 kHz, 12.5 A
  • AC13, AC14, AC15: Sinusoidal cycling 20 kHz, 12.5 A
  • AC16, AC17, AC18: Sinusoidal cycling 40 kHz, 6.25 A
  • AC19, AC20, AC21: Artificial ripple cycling
  • AC22, AC23, AC24: Realistic ripple cycling
  • AC25, AC26, AC27: Realistic ripple cycling

Current Ripple Evaluation

The evaluation results of current ripples in a battery-electric vehicle are stored in the “current_ripple_evaluation” directory. It contains the following files:

  • input_sWLTP.csv/input_UDDS.csv: The speed, torque, and power of the sWLTP and UDDS cycles at every second of the simulated battery-electric vehicle.
  • cycler_sWLTP.csv/cycler_UDDS.csv: The compressed current values for every second of the sWLTP and UDDS cycles. More details about the compression can be found at [1].
  • parameters.csv: The parameters for simulation of the battery-electric vehicle and the drivetrain.
  • sWLTP.h5/UDDS.h5: The simulated current and voltage of the battery-electric vehicle in the time-domain sampled with 500 kHz. Every second has its operating point, which is simulated for a second.


Cell Cycles

The raw current, voltage, and temperature measurements of the cycled battery cells. The results are stored in the “cell_cycling_*” directory with the following files:


  • cell_cycling_sinusoidal/[cell id]/[cell id]_cycle[cycle id].hdf5: The raw measurements of the sinusoidal cycling experiments.
  • cell_cycling_artificial_ripple/[cell id]/[cell id]_cycle[cycle id].hdf5: The raw measurements of the artifical ripple cycling experiments.
  • cell_cycling_realistic_ripple/[cell id]/[cell id]_cycle[cycle id].hdf5: The raw measurements of the realistic ripple cycling experiments.

The measurements have been taken periodically by an external 2MHz measurement system. Each .hdf5 file contains the measurements of a specific cycle number of the experiment. In the sinusoidal and artificial ripple tests, a measurement has been taken for every 1 % State-of-Charge. In the realistic ripple tests, a measurement has been taken every 5 seconds. Every measurement has a duration of 100 ms and a sampling rate of 2 MS/s.

Each measurement is a group in the hdf5 file, with the voltage and current as 1d 32bit floating-point arrays. Each measurement also has a Unix UTC timestamp of the time of the measurement, the cell temperature, capacity, and resistance[10s] stored as attributes. The capacity and resistance are synchronized with the checkups and interpolated linearly over the number of cycles.


Cell Checkups

The periodic checkups of the battery cells are composed of capacity, internal resistance, EIS, OCV, and qOCV measurements. Measurements that take longer, such as EIS and OCV, are not taken at every checkup. The results are stored in the “cell_checkups” directory with the following files:


  • Overview.csv: List of every checkup of every battery cell. For every checkup, the date, number of cycles, capacity, 10s resistance, and references to the more detailed checkup files are stored.
  • EIS/[cell id]_CheckUp[checkup id]_[date]_EIS.csv: The results of the electrochemical impedance spectroscopy from 0.001 to 50,000 Hz using an EIS-meter.
  • OCV/[cell id]_CheckUp[checkup id]_[date]_OCV.csv: Results of the open-circuit voltage measurement between 0 and 100 % State of Charge in 5 % steps.
  • qOCV/[cell id]_CheckUp[checkup id]_[date]_qOCV.csv: Results of the quasi-open-circuit voltage measurement between 0 and 100 % State of Charge in 1 % steps. It contains the quasi-open-circuit voltage, 1s resistance, and 10s resistance.
  • OCV_raw/[cell id]_CheckUp[checkup id]_[date]_OCV_raw.csv: Raw current, voltage, and temperature values of the OCV measurement.
  • qOCV_raw/[cell id]_CheckUp[checkup id]_[date]_qOCV_raw.csv: Raw current, voltage, and temperature values of the qOCV measurement.
  • Capacity_raw/[cell id]_CheckUp[checkup id]_[date]_Cap_raw.csv: Raw current, voltage, and temperature values of the capacity measurement.

The dataset is divided into two parts. The measurement dataset and simulation dataset. The measurement dataset contains received power measurements at 28 GHz in an indoor corridor and outdoor open area. The received power and other channel statistics, e.g., root mean square delay spread, power and time of arrival of multipath components, and path loss were obtained using the PXI channel sounder system. Two different gain antennas 17 dBi and 23 dBi were used. The transmitter was fixed, whereas the receiver was moved in a straight line aligned to the boresight of the transmitter antenna.


The measurement dataset consists of reflected received power from different shaped and sized metallic reflectors at 28 GHz in the indoor corridor and outdoor open area. PXI channel sounder from National Instruments was used for measurements. Horn antennas of gain 17 dBi were used at the transmitter and receiver. The measurements consisted of three flat square metallic reflectors of sizes 0.84 × 0.84 m^2 , 0.61 × 0.61 m^2 , and 0.3 × 0.3 m^2 , a sphere, and a cylinder. The effect of size and shape of the reflectors on the coverage was analyzed in the indoor corridor and outdoor open area.


Datasets for image and video aesthetics

1. Video Dataset : 107 videos This dataset has videos that can be framed into images.

Color contrast,Depth of Field[DoF],Rule of Third[RoT] attributes

that affect aesthetics can be extracted from the video datasets.


2.Slow videos and Fast videos can be assessed for motion

affecting aesthetics


The Firearm Recoil Dataset was collected utilizing a wrist worn accelerometer to record the recoil generated from one subject’s use of 15 different firearms of the Handgun, Rifle and Shotgun class. The type of the firearm based on its ability to auto-load or not is also denoted. 


Datasets are broken up into seperate CSV files for each individual firearm. Details associated with the firearm utilized and data collection specifications is outlined in the Readme File. If you use this datasets for your research, please cite the following paper:

Md. Abdullah Al Hafiz Khan, David Welsh, and Nirmalya Roy. Firearm Detection Using Wrist Worn Tri-Axis Accelerometer Signals, in Proceedings of the 4th Workshop on sensing systems and applications using wrist worn smart devices (WristSense’18), co-located with PerCom, March 2018