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.
156
datasets available to search
ShareScore release 0.9.0
Dataset results
156 results for “Video dataset”
Human embryo time-lapse video dataset - Part 2
<p>A human embryo time-lapse video dataset.</p> <p> </p> <p> </p>
Human embryo time-lapse video dataset - Part 3
<p>A human embryo time-lapse video dataset.</p> <p> </p> <p> </p>
Lightweight Faceswap Video Dataset
<p>This dataset consists of the original 300 videos modified using faceswap manipulation techniques. Faceswap is a lightweight approach that extracts the face region of one image and places it into another. This manipulation method can be run on smartphones.</p>
Original Video Dataset
<p>Since news speakers and journalists tend to be targets of video manipulation, this dataset consists of 300 short videos of news speakers and journalists of various backgrounds. All of the videos are completely pristine (unaltered). </p>
Realistic Large-Scale Fine-Depth Dehazing Dataset from 3D Videos
<p>Dehaze Dataset LSFD</p>
Rural Route Nomad Photo and Video Collection Dataset
<p>This dataset encompasses the metadata drawn from preserving and visualizing the Rural Route Nomad Photo and Video Collection. The collection consists of 14,058 born-digital objects shot on over a dozen digital cameras in over 30 countries, on seven continents from the end of 2008 through 2009. Metadata was generated using ExifTool, along with manual means, utilizing OpenRefine and Excel to parse and clean.</p> <p>The dataset was a result of an overriding project to preserve the digital content of the Rural Route Nomad Collection, and then visualize photographic specs and geographic details with charts, graphs and maps in Tableau. A description of the project as a whole is publicly forthcoming. Visualizations can be found at https://public.tableau.com/app/profile/alan.webber5364.</p>
Dataset: VanEck Video Gaming and eSports ETF (ESPO) Stock Performance
This dataset provides historical stock market performance data for specific companies. It enables users to analyze and understand the past trends and fluctuations in stock prices over time. This information can be utilized for various purposes such as investment analysis, financial research, and market trend forecasting.
VISEM: A Multimodal Video Dataset of Human Spermatozoa
<p>Real multimedia datasets that contain more than just images or text are rare. Even more so are open multimedia datasets in medicine. Often, clinically related datasets only consist of image or videos. We present a dataset that is novel in two ways. Firstly, it is a multi-modal dataset containing different data sources such as videos, biological analysis data, and participant data. Secondly, it is the first dataset of that kind in the field of human reproduction. It consists of anonymized data from 85 different participants. We hope this dataset will inspire people to apply their knowledge in this important field, generate shareable results in the domain, and ultimately improve human infertility investigation and treatment.</p>
Datasets for Video Synopsis
<p>several videos for synopsis research.</p>
Acoustic video cameras multi-species multi-cameras Validation Dataset (VD) for Deep Learning applications
<p>This video dataset, called also VD (Validation Dataset), is designed to test/validate, on a real-world case, deep learning models to identify fish species in sonar camers video flux. It includes data from two different type of cameras (ARIS and DIDSON), two sites (Touques and Selune rivers in Normandy, France), 6 different fishes classes (Atlantic Salmon, European Eel, Sea Lamprey, Allis Shad, European Catfish and generic unidentified fish). This dataset is composed by around 40h of videos, to test the efficiency of the models in the frame of ecological studies and to assess their real-applicability on monitoring sites data. Two sheets are given as the ground truth in which all fish passages (for fish sizes larger than 20 cm) are annotated. No bounding boxes are given.</p>
Subjective Test Dataset and Meta-data-based Models for 360° Streaming Video Quality
<p>During the last years, the number of 360° videos available for streaming has rapidly increased, leading to the<br> need for 360° streaming video quality assessment. In this paper, we report and publish results of three subjective 360° video<br> quality tests, with conditions used to reflect real-world bitrates and resolutions including 4K, 6K and 8K, resulting in 64 stimuli<br> each for the first two tests and 63 for the third. As playout device we used the HTC Vive for the first and HTC Vive Pro<br> for the remaining two tests. Video-quality ratings were collected using the 5-point Absolute Category Rating scale. The 360°<br> dataset provided with the paper contains the links of the used source videos, the raw subjective scores, video-related meta-data,<br> head rotation data and Simulator Sickness Questionnaire results per stimulus and per subject to enable reproducibility of the<br> provided results. Moreover, we use our dataset to compare the performance of state-of-the-art full-reference quality metrics such<br> as VMAF, PSNR, SSIM, ADM2, WS-PSNR and WS-SSIM. Out of all metrics, VMAF was found to show the highest correlation<br> with the subjective scores. Further, we evaluated a center-cropped version of VMAF ("VMAF-cc") that showed to provide a similar<br> performance as the full VMAF. In addition to the dataset and the objective metric evaluation, we propose two new video-quality<br> prediction models, a bitstream meta-data-based model and a hybrid no-reference model using bitrate, resolution and pixel<br> information of the video as input. The new lightweight models provide similar performance as the full-reference models while<br> enabling fast calculations.</p>
KuaiRand: An Unbiased Sequential Recommendation Dataset with Randomly Exposed Videos
<p>The details can be referred to: <a href="https://kuairand.com/" target="_blank" rel="noopener"><strong>https://kuairand.com/</strong></a></p> <p>If it helps you, please kindly cite:</p> <blockquote> <pre><code>@inproceedings{gao2022kuairand, title = {KuaiRand: An Unbiased Sequential Recommendation Dataset with Randomly Exposed Videos}, author = {Gao, Chongming and Li, Shijun and Zhang, Yuan and Chen, Jiawei and Li, Biao and Lei, Wenqiang and Jiang, Peng and He, Xiangnan}, url = {https://doi.org/10.1145/3511808.3557624}, doi = {10.1145/3511808.3557624}, booktitle = {Proceedings of the 31st ACM International Conference on Information and Knowledge Management}, series = {CIKM '22}, location = {Atlanta, GA, USA}, numpages = {5}, year = {2022}, pages = {3953–3957} }</code></pre> </blockquote>
[LS2N_IPI_Salient360] A dataset of head and eye movements for 360° videos
<h1>Datasets & Toolbox</h1> <div> <p>The following datasets and tools have been made available to those interested in developing and benchmarking their models:</p> <ul> <li><a href="https://salient360.ls2n.fr/datasets/training-dataset/"><strong>Training dataset</strong></a>: A dataset containing 360º images and videos and their corresponding ground-truth saliency maps and scan-paths (according to the different types of models), so you can train and tune your algorithms as necessary, and may also compute the benchmark scores as a reference for yourself.</li> <li><a href="https://salient360.ls2n.fr/datasets/toolbox/"><strong>Toolbox</strong></a>: Scripts to parse the provided data and to compute metrics for comparing saliency maps and scan-paths to assess the performance of the models</li> </ul> <p> </p> <h2>The dataset is structured as follows:</h2> <ul> <li>Stimuli: 19 omnidirectional videos of 20 seconds in equi-rectangular format, 85 omnidirectional images in equi-rectangular format.</li> <li>H: Folder containing the saliency maps and scanpaths from head-only movements.</li> <li>HE: Folder containing the saliency maps and scanpaths from head and eye movements.</li> <li>Tools: Python scripts to parse the saliency-map binary files, and to compute saliency and scanpanth measures.</li> </ul> <p> </p> <p>The details about the saliency map files and the scanpath files are:</p> <ul> <li>Saliency maps from head-only movements: Binary files representing the saliency-map sequences are provided. These sequences contain one saliency map per frame with a resolution of 2048x1024. In a binary file, the saliency values (float32) are organized row-wise and one frame after the other. For each sampled head position, the center of the viewport is considered. Then, an isotropic 3.34-degree Gaussian foveation filter centered in the view-port is applied.</li> <li>Scanpaths from head-only movements: Text files are provided with scanpaths from head movement with 100 samples per observer. Each line contains a vector that indicates the fixation index, longitude, latitude and fixation timestamp, respectively. The fixation index is incremented serially for a particular observer and resets to 0 when we reach the next observer, after all of the fixations of the given observer are reported. The fixation starting time is indicated in seconds, and latitude and longitude positions are normalized between 0 and 1 (so they should be multiplied according to the resolution of the desired equi-rectangular image output dimension).</li> <li>Saliency maps from head and eye movements: Binary files representing the saliency-map sequences. These sequences contain one saliency map per frame with a resolution of 2048x1024. In a binary file, the saliency values (float32) are organized row-wise and one frame after the other. For each eye fixation, an isotropic 2-degree gaussian foveation filter centered at the fixation position is applied. This process is applied to the fixations from both left and right eyes, and then combined in the final saliency map.</li> <li>Scanpaths from head and eye movements: Text files are provided with the scanpaths from both left and right eyes. Each line contains a vector that indicates the fixation index, longitude, latitude and fixation timestamp, duration, start frame and end frame, respectively. The fixation index is incremented serially for a particular observer and resets to 0 when we reach the next observer, after all of the fixations of the given observer are reported. The fixation starting time is indicated in seconds, and latitude and longitude positions are normalized between 0 and 1 (so they should be multiplied according to the resolution of the desired equi-rectangular image output dimension).</li> </ul> <p> </p> <h2><strong>Experiment</strong></h2> <h3>Image</h3> <p>The head mounted display (HMD) Oculus-DK2 was used for this test. It has a frame refresh rate of 75Hz, resolution of 960x1080 per eye and a total viewing angle of 100x100 degrees. The gyroscopic sensors within the device are able to transmit the orientation data at a rate equal to the device frame refresh rate. A small eye-tracking camera from Sensomotoric Instruments (SMI) was integrated into the device and was able to transmit eye-tracking data binocularly at 60Hz.</p> <p>The software setup included a custom build unity software along with the Oculus-DK2 driver version 2.0. The software had a feature to check for calibration accuracy every two minutes and re-calibrated each time if necessary.</p> <p>A total of 63 observers in the age group of 19-52 participated in the test. Observers were tested for visual acuity using the Snellen Test and their dominant eye was also determined using the cardboard technique. </p> <p>To maintain a natural (free-viewing like) gaze pattern, subjects were made to view the scene normally without the need to provide explicit quantitative measurements. They were instructed to watch the scene as normally as possible with a combination of head and eye-movement. Observers were also free to stop the test anytime in case they felt fatigued or had a sensation of vertigo. There were five images used as a training for the observers before starting the actual test.</p> <p>A total of 60 stimuli were shown to the observers in a sequence. Each stimuli lasted for 25 seconds and there was a 5 second gray screen between two stimuli. Every two minutes there was a calibration performed to check the accuracy of the eye-tracker. The test itself lasted for about 35 minutes and the observers had a pause of 5 minutes at the half point of the experiment. The observers were themselves seated comfortably in a turn-chair and were free to rotate the full 360 degrees and also move the chair within the room if necessary. The position of each 360 image was reset to the equirectangular image center at the start of each viewing (irrespective of their position). This was done to ensure that all observers start at the same starting position in the panorama.</p> <h3>Video</h3> <p>360-degree videos were displayed in a VR headset (HTC VIVE) equipped with an SMI eye-tracker. The HTC VIVE headset allows sampling of scenes by approximately 110-degrees horizontal by 110-degrees vertical field of view (1080x1200 pixels per eye) monocularly at 90 frames per second. The eye-tracker samples gaze data at 250Hz with a precision of 0.2 degrees. A custom Unity3D scene was created to display videos. </p> <p>57 participants were recruited (25 women; age 19 to 44, mean: 25.7 years), normal or corrected-to-normal vision was verified and dominant eye of all observers was checked. All 19 videos were observed by all observers for their entire duration (20 seconds).</p> <p>Observers were told to freely explore 360-degrees videos as naturally as possible while wearing a VR headset. Videos were played without audio. In order to let participants safely explore the full 360-degrees field of view, we chose to have them seat in a rolling chair. </p> <p>Participants started exploring omnidirectional contents either from an implicit longitudinal center (0-degrees and center of the equirectangular projection) or from the opposite longitude (180-degrees). Videos were observed in both rotation modalities by at least 28 participants each. We controlled observers starting longitudinal position in the scene by offsetting the content longitudinal position at stimuli onset, making sure participants started exploring 360-degrees scenes at exactly 0-degrees, or 180-degrees of longitude according to the modality. Video order and starting position modalities were cross-randomized for all participants.</p> <p>Observers started the experimentation by an eye-tracker calibration, repeated every 5 videos to make sure that eye-tracker's accuracy does not degrade. the total duration of the test was less than 20 minutes.</p> </div>
x264 and x265 performance on eight videos of the Youtube UGC Dataset
<p>The measurements of 3125 configurations of two video encoders, namely <a href="https://www.videolan.org/developers/x264.html">x264</a> and <a href="https://www.videolan.org/developers/x265.html">x265</a>, on eight different videos of the <a href="https://media.withyoutube.com/">Youtube UGC Dataset</a></p>
Node Tracking paper dataset and example videos
Open the record for dataset details and reuse information.
Paired Sonar / Optical Video Dataset
<p>This dataset was collected for L. Viney's undergraduate engineering honors thesis at UNSW in collaboration with Reach Robotics. <br>It features paired forward facing multibeam sonar and optical video recordings gathered using an Oculus M series sonar mounted on a FUSION ROV in Sydney Harbour. <br>Reach Robotics retains rights and ownership of the data, which is released under a Creative Commons CC BY-NC-SA license.</p> <p>Dataset Contents<br>Sonar Recordings: Proprietary-format sonar files compatible with Oculus Viewer software.<br>Optical Video Recordings: Synchronized optical video files paired with each sonar recording.<br>Bounding Box Labels: A limited number of sessions include bounding box annotations for sonar data, provided as CSV files.</p> <p>Prerequisites<br>To view and analyze the sonar data, download the Oculus Viewer software from Blueprint Subsea:<br>https://www.blueprintsubsea.com/oculus/support</p> <p>Folder Structure<br>The dataset is organized with each session containing both sonar and optical video files. Sessions with annotations also contain CSV files for bounding box labels.</p> <p>dataset/<br>├── session1/<br>│ ├── sonar.oculus<br>│ ├── optical.mp4<br>│ └── labels.csv # Bounding box annotations (if available)<br>├── session2/<br>│ ├── sonar.oculus<br>│ ├── optical.mkv<br>│ └── labels.csv # Only in labeled sessions<br>└── README.md</p> <p><br>Data Annotations<br>Annotations are provided as CSV files in specific sessions. Each CSV file includes bounding box coordinates for objects in the sonar frames, with columns typically structured as:</p> <p>frame_id, object_1_x_min, object_1_y_min, object_1_x_max, object_1_y_max, object_2 ...</p> <p>Usage<br>Sonar Data: Use the Oculus Viewer software to load and view .oculus files.<br>Annotations: Use the CSV files for supervised learning or object detection by overlaying bounding boxes on sonar data.<br>Optical Video: Optical video files are compatible with standard video analysis tools.</p> <p>License<br>This dataset is licensed under the Creative Commons CC BY-NC-SA license. Reach Robotics retains ownership and rights to the data.</p>
Human embryo time-lapse video dataset - Trained models
<p>The trained models for the human embryo time-lapse video dataset.</p>
Spine Surgery Video Observation Study. The Creation of a Benchmark Video (RGB-Depth) Dataset to Investigate the Feasibility of Developing a Markerless Tracking System for Spine Surgery.
ClinicalTrials.gov study NCT06580379. IPD Sharing: NO. Countries: 0. Publications: 0.
Dataset for: "Disturbed YouTube for Kids: Characterizing and Detecting Inappropriate Videos Targeting Young Children"
<p>Dataset for paper: Disturbed YouTube for Kids: Characterizing and Detecting Inappropriate Videos Targeting Young Children</p> <p>The dataset consists of five files:<br> 1. groundtruth_videos.json: This is the ground truth dataset. We have 4797 manually annotated videos (1513 suitable, 929 disturbing, 419 restricted, and 1936 irrelevant). You can distinguish among the different labels by observing the 'classification_label' field.<br> 2. elsagate_related_videos.json: Contains the data for 233K elsagate-related YouTube videos (1K seed and 232K recommended) that were obtained as described in the paper.<br> 3. other_child_related_videos.json: Contains the data for 155K other child-related YouTube videos (2K seed and 153K recommended) that were obtained as described in the paper.<br> 4. random_videos.json: Contains the data for 482K random YouTube videos (8K seed and 474K recommended) that were obtained as described in the paper.<br> 5. popular_videos.json: Contains the data for 11K popular YouTube videos (500 seed and 10.5K recommended) that were obtained between November 18 and November 21, 2018, as described in the paper.</p> <p>For each video in all sets, you can check the predicted label of our classifier by observing the 'prediction' field.</p>
TULIP Dataset (CVPR 2024): Multi-Camera Videos and Clinician Ratings of the MDS-UPDRS Part III Motor Exam for Parkinson's Disease Assessment
<div> <p><strong>TULIP Dataset (Version 1.0.3)</strong></p> </div> <div> <p>** Updates! We include one more subject, so total we have 12 subjects in this dataset.</p> </div> <div> <p><strong>Overview:</strong><br>The TULIP (Three-dimensional Understanding and Learning of Impairments in Parkinson’s) dataset provides high-resolution RGB data from multi-camera setups, supporting research on precision motor assessments for Parkinson’s Disease (PD). Version 1.0.0 features synchronized RGB data from six cameras, capturing multiple angles of PD and healthy participants performing clinically relevant motor tasks. We chose the name TULIP, a nod to the floral emblem of PD research and advocacy, to symbolize our goal for this dataset, to foster transformative new machine learning approaches for PD understanding and treatment.</p> </div> <div>This dataset was published as part of our <a title="https://openaccess.thecvf.com/content/CVPR2024/html/Kim_TULIP_Multi-camera_3D_Precision_Assessment_of_Parkinsons_Disease_CVPR_2024_paper.html" href="https://openaccess.thecvf.com/content/CVPR2024/html/Kim_TULIP_Multi-camera_3D_Precision_Assessment_of_Parkinsons_Disease_CVPR_2024_paper.html" target="_blank" rel="noopener noreferrer">CVPR 2024 Paper.</a></div> <div> </div> <div>Kyungdo Kim, Sihan Lyu, Sneha Mantri, Timothy W. Dunn; <em>TULIP: Multi-camera 3D Precision Assessment of Parkinson's Disease</em>; Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024, pp. 22551-22562</div> <p> </p> <p><strong>Data Contents:</strong><br>RGB videos are organized by subject ID and activity, with synchronized recordings from six camera perspectives per activity. Each video documents motor tasks such as gait and finger tapping, in line with the UPDRS standards for PD, allowing for a detailed study of joint angles, tremors, and other movements key to understanding PD progression.</p> <p>· <em><strong> </strong>Multi-Camera Video Data:</em> RGB videos from six cameras enable robust 3D pose extraction.</p> <p>· <em>Metadata: </em>Includes camera parameters (intrinsic and extrinsic matrices for 3D reconstruction) and task descriptions.</p> <div> <p>· <em>Clinical Examination Labels:</em> Task-specific labels aligned with clinical motor assessments, with annotations from three clinicians to aid in automated scoring models. We also provided the labels in the csv file format.</p> </div> <div> <p> </p> </div> <p><strong>File Structure:</strong><br>Data is organized by Subject ID > Activities > Camera Perspective, with each folder containing RGB video files. Annotations and activity labels are available in a CSV file for easy correlation of tasks with motor patterns. For the camera parameters (pickle file), each subject has its own set of parameters. When you open the pickle file, the order of the elements is as follows: [proj_matrices, cam_matrices, extrinsic_matrices, rmatrices, rvecs, tvecs, distcoeffs]. Here’s a brief description of each:</p> <p>· <em>proj_matrices: </em>Camera projection matrix (3x4 format)</p> <p>· <em>cam_matrices:</em> Intrinsic camera matrix</p> <p>· <em>extrinsic_matrices:</em> Extrinsic camera matrix</p> <p>· <em>rmatrices:</em> Rotation matrix</p> <p>· <em>rvecs:</em> Rotation vector</p> <p>· <em>tvecs:</em> Translation vector</p> <div> <p>· <em>distcoeffs:</em> Distortion coefficients, which is a zero matrix in our case.</p> </div> <div> <p> </p> </div> <p><strong>Privacy and Consent:</strong><br>Faces are blurred to ensure privacy. This initial version of the dataset contains data from 12 participants, with data from the remaining 3 participants expected to be released in the near future.</p> <p><strong>Code Availability:</strong><br>Behavioral feature extraction demo code for 3D poses will be available on our github (github link can be found on the <a title="https://www.tulipproject.net/" href="https://www.tulipproject.net/" target="_blank" rel="noopener noreferrer">TULIP Project</a> page). For further details and access to our publication on TULIP data and baseline projects, please visit <a title="https://www.tulipproject.net/" href="https://www.tulipproject.net/" target="_blank" rel="noopener noreferrer">TULIP Project</a> or <a title="https://openaccess.thecvf.com/content/CVPR2024/html/Kim_TULIP_Multi-camera_3D_Precision_Assessment_of_Parkinsons_Disease_CVPR_2024_paper.html" href="https://openaccess.thecvf.com/content/CVPR2024/html/Kim_TULIP_Multi-camera_3D_Precision_Assessment_of_Parkinsons_Disease_CVPR_2024_paper.html" target="_blank" rel="noopener noreferrer">CVPR 2024 Paper</a>.</p> <p><strong>Citing this dataset:</strong></p> <p>Please cite our CVPR paper if use this dataset in your work. Citation:</p> <p>Kyungdo Kim, Sihan Lyu, Sneha Mantri, Timothy W. Dunn; <em>TULIP: Multi-camera 3D Precision Assessment of Parkinson's Disease</em>; Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024, pp. 22551-22562</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.