Find research datasets worth reusing
Search datasets from major research repositories and use ShareScore to quickly assess how well each record supports discovery, access, and reuse.
8
datasets available to search
ShareScore release 0.9.0
Dataset results
8 results for “Event Camera”
Aircraft Marshaling Signals Dataset of FMCW Radar and Event-Based Camera for Sensor Fusion
<p><strong>Dataset Introduction</strong></p><p>The advent of neural networks capable of learning salient features from variance in the radar data has expanded the breadth of radar applications, often as an alternative sensor or a complementary modality to camera vision. Gesture recognition for command control is arguably the most commonly explored application. Nevertheless, more suitable benchmarking datasets than currently available are needed to assess and compare the merits of the different proposed solutions and explore a broader range of scenarios than simple hand-gesturing a few centimeters away from a radar transmitter/receiver. Most current publicly available radar datasets used in gesture recognition provide limited diversity, do not provide access to raw ADC data, and are not significantly challenging. To address these shortcomings, we created and make available a new dataset that combines FMCW radar and dynamic vision camera of 10 aircraft marshalling signals (whole body) at several distances and angles from the sensors, recorded from 13 people. The two modalities are hardware synchronized using the radar's PRI signal. Moreover, in the supporting publication we propose a sparse encoding of the time domain (ADC) signals that achieve a dramatic data rate reduction (>76%) while retaining the efficacy of the downstream FFT processing (<2% accuracy loss on recognition tasks), and can be used to create an sparse event-based representation of the radar data. In this way the dataset can be used as a two-modality neuromorphic dataset.</p><p><strong>Synchronization of the two modalities</strong></p><p>The PRI pulses from the radar have been hard-wired to the event stream of the DVS sensor, and timestamped using the DVS clock. Based on this signal the DVS event stream has been segmented such that groups of events (time-bins) of the DVS are mapped with individual radar pulses (chirps).</p><p><strong>Data storage</strong></p><p>DVS events (x,y coords and timestamps) are stored in structured arrays, and one such structured array object is associated with the data of a radar transmission (pulse/chirp). A radar transmission is a vector of 512 ADC levels that correspond to sampling points of chirping signal (FMCW radar) that lasts about ~1.3ms. Every 192 radar transmissions are stacked in a matrix called a radar frame (each transmission is a row in that matrix). A data capture (recording) consisting of some thousands of continuous radar transmissions is therefore segmented in a number of radar frames. Finally radar frames and the corresponding DVS structured arrays are stored in separate containers in a custom-made multi-container file format (extension .rad). We provide a (rad file) parser for extracting the data out of these files. There is one file per capture of continuous gesture recording of about 10s.</p><p>Note the number of 192 transmissions per radar frame is an ad-hoc segmentation that suits the purpose of obtaining sufficient signal resolution in a 2D FFT typical in radar signal processing, for the range resolution of the specific radar. It also served the purpose of fast streaming storing of the data during capture. For extracting individual data points for the dataset however, one can pool together (concat) all the radar frames from a single capture file and re-segment them according to liking. The data loader that we provide offers this, with a default of re-segmenting every 769 transmissions (about 1s of gesturing).</p><p><strong>Data captures directory organization (</strong><a href="https://zenodo.org/api/records/10359770/draft/files/radar8Ghz-DVS-marshaling_signals_20220901_publication_anonymized.7z/content">radar8Ghz-DVS-marshaling_signals_20220901_publication_anonymized.7z</a><strong>)</strong></p><p>The dataset captures (recordings) are organized in a common directory structure which encompasses additional metadata information about the captures.</p><p>dataset_dir/<stage>/<room>/<person>-<gesture>-<distance>/ofxRadar8Ghz_yyyy-mm-dd_HH-MM-SS.rad</p><p>Identifiers</p><ul><li>stage [train, test].</li><li>room: [conference_room, foyer, open_space].</li><li>subject: [0-9]. Note that 0 stands for no person, and 1 for an unlabeled, random person (only present in test).</li><li>gesture: ['none', 'emergency_stop', 'move_ahead', 'move_back_v1', 'move_back_v2', 'slow_down' 'start_engines', 'stop_engines', 'straight_ahead', 'turn_left', 'turn_right'].</li><li>distance: ['xxx', '100', '150', '200', '250', '300', '350', '400', '450'] (in cm). Note that xxx is used for none gestures when there is no person present in front of the radar (i.e. background samples), or when a person is walking in front of the radar with varying distances but performing no gesture.</li></ul><p>The test data captures contain both subjects that appear in the train data as well as previously <i>unseen</i> subjects. Similarly the test data contain captures from the spaces that train data were recorded at, as well as from a new <i>unseen</i> open space.</p><p><strong>Files List</strong></p><p><a href="https://zenodo.org/api/records/10359770/draft/files/radar8Ghz-DVS-marshaling_signals_20220901_publication_anonymized.7z/content">radar8Ghz-DVS-marshaling_signals_20220901_publication_anonymized.7z</a></p><p>This is the actual archive bundle with the data captures (recordings).</p><p><a href="https://zenodo.org/api/records/10359770/draft/files/rad_file_parser_2.py/content">rad_file_parser_2.py</a></p><p>Parser for individual .rad files, which contain capture data.</p><p><a href="https://zenodo.org/api/records/10359770/draft/files/loader.py/content">loader.py</a></p><p>A convenience PyTorch Dataset loader (partly Tonic compatible). You practically only need this to quick-start if you don't want to delve too much into code reading. When you init a DvsRadarAircraftMarshallingSignals class object it automatically downloads the dataset archive and the .rad file parser, unpacks the archive, and imports the .rad parser to load the data. One can then <i>request from it </i>a training set, a validation set and a test set as torch.Datasets to work with<i>.</i> </p><p><a href="https://zenodo.org/api/records/10359770/draft/files/aircraft_marshalling_signals_howto.ipynb/content">aircraft_marshalling_signals_howto.ipynb</a></p><p>Jupyter notebook for exemplary basic use of loader.py</p><p><strong>Contact</strong></p><p>For further information or questions try contacting first M. Sifalakis or F. Corradi.</p><p> </p>
The INI-30 Dataset : Event Camera for Eye Tracking
<p>The Ini-30 dataset is collected with two event cameras mounted on a glass frame. Each DVXplorer sensor (640 × 480 pixels) is attached on the side of the frame. The power supply was provided via a 2 meter cable connected from the cameras to a computer, which provided enough freedom of movement. Differently from [2, 24], the participants were not instructed to follow a dot on a screen, but rather encouraged to look around to collect natural eye movements. As shown in Fig. 1, the event cameras were securely screwed on a 3D-printed case attached to the side of the glass frame. The data was annotated based on accumulated linearly decayed events by defining the pixel intensity as function of the linear accumulation of previous pixel intensity. Next we labeled the position of the pupil in the DVS’s array manually, using an assistive labeling tool. We discarded the first 20ms of events to ensure the eye was visible and annotations met the level of image-based annotators. The number of labels per recording was intentionally variable, spanning from 475 to 1’848 with a time per label ranging from 20.0 to 235.77 milliseconds depending on the overall duration of the sample. This setup allows for unconstrained head movements, enables to capture event data from eye movement in a ”in-the-wild” setting and allows the generation of a representative, unique, diverse and challenging dataset.<br><br>NOTE : the annotations relates to the ellipse of the pupil on the image</p>
Simulation Data & R scripts for: "Introducing recurrent events analyses to assess species interactions based on camera trap data: a comparison with time-to-first-event approaches"
<p><strong>Files descriptions:</strong></p> <p>All csv files refer to results from the different models (PAMM, AARs, Linear models, MRPPs) on each iteration of the simulation. One row being one iteration. <br>"results_perfect_detection.csv" refers to the results from the first simulation part with all the observations.<br>"results_imperfect_detection.csv" refers to the results from the first simulation part with randomly thinned observations to mimick imperfect detection.</p> <p>ID_run: identified of the iteration (N: number of sites, D_AB: duration of the effect of A on B, D_BA: duration of the effect of B on A, AB: effect of A on B, BA: effect of B on A, Se: seed number of the iteration).<br>PAMM30: p-value of the PAMM running on the 30-days survey.<br>PAMM7: p-value of the PAMM running on the 7-days survey.<br>AAR1: ratio value for the Avoidance-Attraction-Ratio calculating AB/BA.<br>AAR2: ratio value for the Avoidance-Attraction-Ratio calculating BAB/BB.<br>Harmsen_P: p-value from the linear model with interaction Species1*Species2 from Harmsen et al. (2009).<br>Niedballa_P: p-value from the linear model comparing AB to BA (Niedballa et al. 2021).<br>Karanth_permA: rank of the observed interval duration median (AB and BA undifferenciated) compared to the randomized median distribution, when permuting on species A (Karanth et al. 2017).<br>MurphyAB_permA: rank of the observed AB interval duration median compared to the randomized median distribution, when permuting on species A (Murphy et al. 2021). <br>MurphyBA_permA: rank of the observed BA interval duration median compared to the randomized median distribution, when permuting on species A (Murphy et al. 2021). <br>Karanth_permB: rank of the observed interval duration median (AB and BA undifferenciated) compared to the randomized median distribution, when permuting on species B (Karanth et al. 2017).<br>MurphyAB_permB: rank of the observed AB interval duration median compared to the randomized median distribution, when permuting on species B (Murphy et al. 2021). <br>MurphyBA_permB: rank of the observed BA interval duration median compared to the randomized median distribution, when permuting on species B (Murphy et al. 2021). <br> </p> <p>"results_int_dir_perf_det.csv" refers to the results from the second simulation part, with all the observations.<br>"results_int_dir_imperf_det.csv" refers to the results from the second simulation part, with randomly thinned observations to mimick imperfect detection.<br>ID_run: identified of the iteration (N: number of sites, D_AB: duration of the effect of A on B, D_BA: duration of the effect of B on A, AB: effect of A on B, BA: effect of B on A, Se: seed number of the iteration).<br>p_pamm7_AB: p-value of the PAMM running on the 7-days survey testing for the effect of A on B.<br>p_pamm7_AB: p-value of the PAMM running on the 7-days survey testing for the effect of B on A.<br>AAR1: ratio value for the Avoidance-Attraction-Ratio calculating AB/BA.<br>AAR2_BAB: ratio value for the Avoidance-Attraction-Ratio calculating BAB/BB.<br>AAR2_ABA: ratio value for the Avoidance-Attraction-Ratio calculating ABA/AA.<br>Harmsen_P: p-value from the linear model with interaction Species1*Species2 from Harmsen et al. (2009).<br>Niedballa_P: p-value from the linear model comparing AB to BA (Niedballa et al. 2021).<br>Karanth_permA: rank of the observed interval duration median (AB and BA undifferenciated) compared to the randomized median distribution, when permuting on species A (Karanth et al. 2017).<br>MurphyAB_permA: rank of the observed AB interval duration median compared to the randomized median distribution, when permuting on species A (Murphy et al. 2021). <br>MurphyBA_permA: rank of the observed BA interval duration median compared to the randomized median distribution, when permuting on species A (Murphy et al. 2021). <br>Karanth_permB: rank of the observed interval duration median (AB and BA undifferenciated) compared to the randomized median distribution, when permuting on species B (Karanth et al. 2017).<br>MurphyAB_permB: rank of the observed AB interval duration median compared to the randomized median distribution, when permuting on species B (Murphy et al. 2021). <br>MurphyBA_permB: rank of the observed BA interval duration median compared to the randomized median distribution, when permuting on species B (Murphy et al. 2021). <br> </p> <p><strong>Scripts files description:</strong><br>1_Functions: R script containing the functions:<br> - MRPP from Karanth et al. (2017) adapted here for time efficiency.<br> - MRPP from Murphy et al. (2021) adapted here for time efficiency.<br> - Version of the ct_to_recurrent() function from the recurrent package adapted to process parallized on the simulation datasets.<br> - The simulation() function used to simulate two species observations with reciprocal effect on each other.<br>2_Simulations: R script containing the parameters definitions for all iterations (for the two parts of the simulations), the simulation paralellization and the random thinning mimicking imperfect detection.<br>3_Approaches comparison: R script containing the fit of the different models tested on the simulated data.<br>3_1_Real data comparison: R script containing the fit of the different models tested on the real data example from Murphy et al. 2021.<br>4_Graphs: R script containing the code for plotting results from the simulation part and appendices.<br>5_1_Appendix - Check for similarity between codes for Karanth et al 2017 method: R script containing Karanth et al. (2017) and Murphy et al. (2021) codes lines and the adapted version for time-efficiency matter and a comparison to verify similarity of results.<br>5_2_Appendix - Multi-response procedure permutation difference: R script containing R code to test for difference of the MRPPs approaches according to the species on which permutation are done.</p>
Insights into Batch Selection for Event-camera Motion Estimation - Datasets
<p>See <a href="https://github.com/event-driven-robotics/batch-selection-experiments">https://github.com/event-driven-robotics/batch-selection-experiments</a> for examples on how to read and process the data.</p> <p>Simulated datasets were used to accurately have a ground-truth of 3 DoF camera rotation, for valid training and evaluation. Four photo-realistic simulated environments from UnrealCV Zoo were selected as they are publicly available indoor scenes and include diverse lighting textures, shadows, reflections and object clutter. The camera was positioned inside the virtual room and rotated along all its three axes randomly at a variety of speeds. Frames were generated at >1 kHz from which the event-stream is generated using log-image-difference techniques. Five velocity trajectories were used for <em>arch1</em>, four velocity trajectories were used for <em>arch2</em>, and three trajectories for <em>arch3</em>. Including the final <em>room</em> dataset, a total of 13 different datasets (each with a different velocity trajectory) cover a total of 190 seconds of data. Note: <em>room</em> is split into 3 archived files.</p> <p>Please cite:</p> <blockquote> <p>Valerdi, J.L., Bartolozzi, C. and Glover, A., 2023. Insights into Batch Selection for Event-Camera Motion Estimation. <em>Sensors</em>, <em>23</em>(7), p.3699.</p> </blockquote> <pre><code>@article{valerdi2023insights, title={Insights into Batch Selection for Event-Camera Motion Estimation}, author={Valerdi, Juan L and Bartolozzi, Chiara and Glover, Arren}, journal={Sensors}, volume={23}, number={7}, pages={3699}, year={2023}, publisher={MDPI} }</code></pre>
High-throughput event-based and frame-based convolutions for event-cameras
<p>Event cameras are promising sensors for on-line and real-time vision tasks, due to their high temporal resolution, low latency and the elimination of redundant static data. Many vision algorithms use some form of spatial convolution (i.e. spatial pattern detection) as a fundamental component, but additional consideration must be taken for event cameras, as the visual signal is asynchronous and sparse. While elegant methods have been proposed for event-based convolutions, they are unsuitable for real scenarios due to their inefficient processing pipeline, and subsequent low event-throughput. This paper presents an efficient implementation based on decoupling the event-based computations from the computationally heavy convolution ones, increasing the maximum event processing rate by 15.92x, to over 10 million events/second, while still maintaining the event-based paradigm of asynchronous input and output. Results on public datasets with modern 640x480 event-camera recordings show that the proposed implementation achieves real-time processing with minimal impact in the convolution result, while the prior state-of-the-art results in latency of over 1 second per-event.</p>
Neuromorphic sequence learning with an event camera on routes through vegetation
<p>code and dataset for paper 'Neuromorphic sequence learning with an event camera on routes through vegetation'.</p>
EDOPT: Event-camera 6-DoF Dynamic Object Pose Tracking
<p>The dataset can be used to test your event-based 6-DoF pose tracking algorithm.</p> <p>If you use any of this data, please cite the following publication:</p> <p>@inproceedings{glover2024,<br> title={EDOPT: Event-camera 6-DoF Dynamic Object Pose Tracking },<br> author={Glover, Arren and Gava, Luna and Li, Zhichao and Bartolozzi, Chiara},<br> booktitle={2024 IEEE International Conference on Robotics and Automation (ICRA)},<br> year={2024}<br>}</p> <p>The dataset includes event-driven data and ground truth of 5 objects: dragon, jell-o, mustard, soup can, and spam. For each object, six different motions on independent axes are recorded. </p> <p>To import .log files containing events, we suggest <a href="https://github.com/event-driven-robotics/bimvee">bimvee</a> Python library.</p> <p>Specifically, use the functions to import .log files:</p> <p>data = importIitYarp(filePathOrName=input_path)</p> <p>Ground-truth .csv files have 8 columns, each one corresponding to a different measure: </p> <p>timestamp | x | y | z | qx | qy | qz | qw</p> <p>x, y, z refer to the object position in the camera reference frame, while qx, qy, qz and qw refer to the object orientation expressed in quaternions. </p>
Fast Trajectory End-Point Prediction with Event Cameras for Reactive Robot Control
<p>If you use any of this data, please cite the following publication:</p> <p><span>@inproceedings{monforte2023fast,</span><br><span> title={Fast Trajectory End-Point Prediction with Event Cameras for Reactive Robot Control},</span><br><span> author={Monforte, Marco and Gava, Luna and Iacono, Massimiliano and Glover, Arren and Bartolozzi, Chiara},</span><br><span> booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},</span><br><span> pages={4035--4043},</span><br><span> year={2023}</span><br><span>}</span></p> <p>Event-based datasets of synthetic and real trajectories of a bouncing ball.</p> <p>The synthetic trajectories were obtained converting frames taken using Unreal Engine to events. The ground truth is provided along with objects and camera settings.</p> <p>The real trajectories wer dumped from a real event camera located in front of the robot workspace.</p> <p>To import .log files containing events, we suggest <a href="https://github.com/event-driven-robotics/bimvee">bimvee</a> Python library.</p> <p>Specifically use the functions to import .log files:</p> <p>data = importIitYarpBinaryDataLog(filePathOrName=input_path)<br> </p> <p> </p> <p> </p>
ScienceDex guides
Understand access before you commit
These curated guides explain access requirements, typical timelines, costs, and reuse considerations for widely used research datasets.
Allen Brain Atlas
Allen Brain Atlas is an Allen Institute collection of brain map atlases, datasets, APIs, and analysis tools covering mouse, human, and non-human primate brain resources.
Annotated Behaviour and Observability Dataset (ABODe)
ABODe is a University of Edinburgh DataShare dataset for behavior classification in group-housed mice using home-cage video, identities, bounding boxes, ground-plate positions, and annotator labels.
DANDI Archive for NWB datasets
DANDI is a BRAIN Initiative archive for publishing and sharing neurophysiology data, including electrophysiology, optophysiology, and behavioral data packaged as NWB and related standards.
International Brain Laboratory public data
The International Brain Laboratory public data releases expose standardized mouse decision-making experiments, including Neuropixels recordings, widefield calcium imaging, behavior, and session metadata accessed through the ONE API.
OpenNeuro
OpenNeuro is a free, open platform for sharing neuroimaging datasets, with public search, dataset pages, and download paths for web, S3, DataLad, and the OpenNeuro CLI.