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.
107
datasets available to search
ShareScore release 0.7.1
Dataset results
107 results for “computer vision”
Super-resolving ocean dynamics from space with computer vision algorithms: training datasets
<p>We provide here the datasets used for the development of the dilated Adaptive Residual Network for the super-resolution of ocean Absolute Dynamic Topography described in <em>Buongiorno Nardelli et al.</em> (2022). The model is designed to combine satellite altimetry and thermal observations and provides super-resolved dynamic topography. The training/test datasets have been built starting from the data originally prepared for an Observing System Simulation Experiment carried out in the framework of the European Space Agency CIRCOL project [<em>Ciani et al.</em>, 2021]. They consist of one year of synthetic daily Absolute Dynamic Topography (ADT), surface geostrophic currents and sea surface temperature data obtained from Copernicus Marine Service Mediterranean Forecasting System (MFS) (Product ID: MEDSEA-ANALYSIS- FORECAST-PHY-006-013) [<em>Clementi et al. 2021</em>]. Synthetic Altimeter-derived ADT maps were obtained by first sampling the model output along the actual tracks of a synthetic constellation composed of 4 Radar Altimeters: Jason-3, Sentinel-3A, SARAL/Altika, and Cryosat-2 missions (this step is achieved by running the SWOT simulator software [<em>Gaultier et al.</em>, 2016]) and successively applying the DUACS (<em>Data Unification and Altimeter Combination System)</em> mapping method. The original input images cover the entire Mediterranean domain at 1/24° spatial resolution, leading to an individual image size of 380x1000 pixels. Here, we have randomly chosen 40 dates (~11% of the total) to be kept aside as fully independent test data, and successively re-sampled the original images extracting much smaller tiles (76x100), which are used as input to the network training. The tiles are extracted by going through a double loop on latitude and longitude, imposing a spatial overlap of 50%. Full details on data pre-processing (e.g.normalization strategies) are given in the paper:</p> <ul> <li>Buongiorno Nardelli, B.; Cavaliere, D.; Charles, E.; Ciani, D. Super-Resolving Ocean Dynamics from Space with Computer Vision Algorithms. <em>Remote Sens.</em>, <strong>2022</strong>, 14, 1159. https://doi.org/10.3390/rs14051159</li> </ul>
Image Databases for Computer Vision Coded for Subject Traceability
<p>This document consists of the corpus of image databases examined for traceability of dataset subjects as published in:</p> <p>Morgan Klaus Scheuerman, Katy Weathington, Tarun Mugunthan, Emily Denton, and Casey Fiesler. 2023. From Human to Data to Dataset: Mapping the Traceability of Human Subjects in Computer Vision Datasets. Proc. ACM Hum.-Comput. Interact. 7, CSCW1, Article 55 (April 2023), 33 pages. https://doi.org/10.1145/3579488</p>
MCR LTER: Coral Reef: Computer Vision: Moorea Labeled Corals
The Moorea Labeled Corals dataset is a subset of the MCR LTER packaged for computer vision research. It contains 2055 images from three habitats IDs: fringing reef outer 10m and outer 17m, from 2008, 2009 and 2010. It also contains random point annotation (row, col, label) for the nine most abundant labels, four non coral labels: (1) Crustose Coralline Algae (CCA), (2) Turf algae, (3) Macroalgae and (4) Sand, and five coral genera: (5) Acropora, (6) Pavona, (7) Montipora, (8) Pocillopora, and (9) Porites. These nine classes account for 96% of the annotations and total to almost 400,000 points. These nine classes are the ones analyzed in (Beijbom, 2012); less-abundant genera not treated in the automation are also present in the dataset. These data were published in Beijbom O., Edmunds P.J., Kline D.I., Mitchell G.B., Kriegman D., 'Automated Annotation of Coral Reef Survey Images', IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Providence, Rhode Island, 2012. [BibTex] [pdf] These data are a subset of the raw data from which knb-lter-mcr.4 is derived. This material is based upon work supported by the U.S. National Science Foundation under Grant No. OCE 16-37396 (and earlier awards) as well as a generous gift from the Gordon and Betty Moore Foundation. Research was completed under permits issued by the French Polynesian Government (Délégation à la Recherche) and the Haut-commissariat de la République en Polynésie Francaise (DTRT) (Protocole d'Accueil 2005-2018). This work represents a contribution of the Moorea Coral Reef (MCR) LTER Site.
High-throughput robotic titration using computer vision
<ul> <li> <p>An automated HTE robotic titration using a liquid-handling robot Opentrons(OT-2) and a standard webcam enables in-situ, affordable titration analyses.</p> </li> <li>Its modular design allows adaptability for materials chemsitry and integration into automated workflows, enhancing efficiency in chemical search.</li> </ul>
EOL computer vision pipelines: Classification for Image Tagging: Flower Fruit
<p>Angiosperms: Stats from Colab:</p> <ul> <li>Number of positive identified reproductive structures: 490</li> <li>Number of possible identified reproductive structures: 4611</li> <li>Number of negative identified reproductive structures: 14833</li> </ul> <p> </p>
EOL computer vision pipelines: Classification for Image Tagging: Image Type: Anura
<p>Produced by the EOL Image Type Classifier. Classifies images as map, phylogeny, illustration, herbarium sheet, or none. Dataset generated for EOL Anura images. See model on the CV for <a href="https://www.kaggle.com/models/eolorg/image-quality-rating-bad-vs-good" target="_blank" rel="noopener">EOL Images Model Zoo on Kaggle</a>.</p>
EOL computer vision pipelines: Classification for Image Tagging: Image Rating: Chiroptera
<p>Produced by the EOL Image Rating Classifier. Classifies images as bad or good quality (used for image gallery sorting). Dataset generated for EOL Chiroptera images. See model on the CV for <a href="https://www.kaggle.com/models/eolorg/image-quality-rating-bad-or-good" target="_blank" rel="noopener">EOL Images Model Zoo on Kaggle</a>.</p>
Computer Vision Datasets for Visual Blockage Assessment at Culverts
<p>Blockage of culverts caused by transported debris is a major factor in causing flash floods in urban areas. Traditional hydraulic models have been unsuccessful in solving this problem due to a lack of data on peak flood hydraulics and the complex behavior of debris at culverts. To address this problem, a new approach of developing intelligent video analytics (IVA) algorithms is being proposed, which uses computer vision algorithms to extract information about visual blockage. This approach is expected to help in timely and safe maintenance operations and reduce the risk of culverts being blocked. To support the development of computer vision solutions, two datasets have been created: the Synthetic Images of Culvert (SIC) and the Visual Hydraulics Lab Dataset (VHD).</p> <ul> <li>The Synthetic Images of Culvert (SIC) dataset consists of synthetic images of culverts that were generated using a 3D computer application built on the Unity3D gaming engine. The application was designed to simulate various blockage scenarios by allowing users to place different types of debris materials in the scene in various orientations and locations. These blockage scenarios were captured as images using batch capture functionality. The dataset offers diversity in terms of the type of debris (urban, vegetative, mixed), culvert types (pipe, single circular, double circular, single box, double box, triple box), camera viewpoints, time of day, and water levels. However, it has some limitations, such as a single natural background and unrealistic effects and animations.</li> <li>The Visual Hydraulics-Lab Dataset (VHD) is a dataset of simulated images of culverts that were captured during controlled hydrology lab experiments. The experiments involved a series of tests using scaled physical models of culverts under different flooding conditions. The experiments were recorded using two high definition (HD) cameras and images of culverts in both blocked and unblocked conditions were extracted. The VHD dataset includes a variety of images with different culvert configurations (single circular, double circular, single box, double box), blockage types (urban, vegetative, mixed), simulated lighting conditions, camera viewpoints, and flood levels controlled by inlet water discharge. The limitations of the dataset include reflections from the water surface and flume walls, an identical background and scaling, and clear water.</li> </ul>
MCR LTER: Coral Reef: Computer Vision: Multi-annotator Comparison of Coral Photo Quadrat Analysis
This repository contains the Moorea portion of a larger data package published in conjuncture with: "Towards automated annotation of benthic survey images: variability of human experts and operational modes of automation", Beijbom et al. PLOS One, 2015. The rest of the data package is hosted at the Dryad data repository (doi:10.5061/dryad.m5pr3). The larger data package is an aggregate dataset from four Pacific coral reef monitoring projects in: Moorea (French Polynesia), the northern Line Islands, Nanwan Bay (Taiwan) and Heron Reef (Australia). It contains 5090 coral reef survey images, and 251,988 random-point annotations by coral ecology experts. The point-annotations indicate the dominant benthic substrate at 10 to 200 random point locations per image, using a label-set of 20 categories. In addition, 200 images from each location have been cross-annotated by 6 experts, for a total of 7 sets of annotations for each image. This set of cross-annotations can be used to contextualize the performance of automated annotation methods for coral reef ecology. The full data package can also be used by computer-vision and machine learning researchers to develop object classification, image segmentation, and domain transfer learning methods. These data contain a subset of the raw data from which dataset knb-lter-mcr.4 is derived.
Search-and-Rescue From Drones With Computer Vision
<p>Unmanned aerial vehicles (UAVs), most commonly known as drones, are increasingly used as a technological support tool for search-and-rescue (SAR) operations (and post-disaster area explorations as well). UAVs equipped with high-resolution cameras and embedded, yet powerful GPUs, in fact, can provide an effective and efficient aid to emergency rescue operations, mainly because locating victims, which may be unconscious or injured, as much fast as possible, is crucial to improve their chance of survival. In particular, the use of drones that are able to automatically detect people in the scenes can increase detection rate, while reducing rescue time. In this repository, we provide a new dataset specifically conceived for SAR operations from drones with computer vision. As it is small-sized, the dataset is currently intended for testing and evaluation purposes only. The main aim of the repository is to encourage contributions on this intriguing topic. In particular, any contribution to make the dataset bigger is welcome.</p>
Cutting the Frame: An In-Depth Look at the Hitchcock Computer Vision Dataset
<p>The Hitchcock Computer Vision Dataset is a comprehensive collection of annotated frames from fifteen Alfred Hitchcock films, created for the purpose of advancing research in film studies and digital humanities. The dataset leverages the power of Google's Vision API to provide detailed annotations such as object detection, facial recognition, web-entity analysis, explicit content filtering, and more. The dataset consists of a CSV file containing about 105,000 frames, uniformly extracted from the films using a time-based approach. Each frame is accompanied by metadata, including the film name, timestamp, and release year, allowing researchers to explore the evolution of Hitchcock's cinematic techniques and recurring themes.</p>
Supplementary Materials for Learning Manufacturing Computer Vision Systems Using Tiny YOLOv4
<h1>About This Dataset</h1> <p>This repository contains the supplementary materials presented in the publication “Learning Manufacturing Computer Vision Systems Using Tiny YOLO v4” by Medina, A., Bradley, R., Xu, W., Ponce, P., Anthony, B., and Molina, A. that can be found with the following DOI <a href="https://www.frontiersin.org/articles/10.3389/frobt.2024.1331249/">10.3389/frobt.2024.1331249</a></p> <p>There are three files in this repository:</p> <ol> <li>dataset.zip</li> <li>YOLOv4_object_detection.ipynb</li> <li>deploy.py</li> </ol> <h1>dataset.zip</h1> <p>This Dataset is for an example used for education purposes. It is a small dataset that is adapted from the following Kaggle repository, authored by Ruthger Righart <a href="https://www.kaggle.com/datasets/rrighart/jarlids/data">https://www.kaggle.com/datasets/rrighart/jarlids/data</a>. One of the activities proposed is to teach students how to find, download and review a free dataset, so this is the example given.</p> <p>Another activity is to teach how to label images to create a custom dataset. The images (with extension .JPG) from the original repository are used. The labels (with extension .txt) were created by the authors of the Learning Manufacturing Computer Vision Systems Using Tiny YOLOv4 paper. The authors used the free tool labelImg, from GitHub repository (<a href="https://github.com/HumanSignal/labelImg">https://github.com/HumanSignal/labelImg</a>), to label the images with object bounding boxes and corresponding labels in the YOLO format.</p> <p>The dataset contains 238 images and corresponding labels, with files named “p<num>.JPG” and “p<num>.txt”. The text labels are formatted in the YOLO format with each row in the .txt file corresponding to one object in the image. Each row contains 5 elements: The object identifier, top left corner x coordinate, top left corner y coordinate, height, and width, separated by a whitespace. The object identifier represents good cans as 0 and defective cans as 1.</p> <h1>YOLOv4_object_detection.ipynb</h1> <p>This notebook was created to give the user a step-by-step tutorial on how to train a YOLOv4 algorithm with a custom dataset using a free GPU on Google Collab, the prerequisite to use it are:</p> <ul> <li>To have ready the dataset.</li> <li>Have the training txt file with the path to all images used for training.</li> <li>Have the test txt file with the path to all images used for testing.</li> </ul> <p>There are other requirements like cloning a GitHub repository and altering certain files on that repository; however, those steps are discussed within the notebook.</p> <p>At the end of the notebook an example on how to test the trained model with images and/or videos is shown, however since Google Collab doesn’t have access to the physical computer of the user live stream video is not part of the example.</p> <h1>deploy.py</h1> <p><em>Disclaimer: This code is not optimized, and its intended purpose is to teach students how to run YOLO on a raspberry pi using the OpenCV library.</em></p> <p>To use this code with different files or datasets, be sure to change the two parameters inside the net3 variable which are the cfg file used while training the algorithm and the weights file. You should also change the class list to include your classes, keeping in mind that the classes order must correspond to the order of the labeling process and class 0 is the first one on the list.</p> <p>Also to change the Title of the created image prompt you shout go to the line calling the imshow method and change the ‘Tiny YOLOv4’ string.</p> <p>This algorithm uses the first camera it finds and opens up a display image with the detected objects surrounded by a bounding box, on top of that box the top predicted class is going to show, to change color of bounding boxes or text change the rectangle method where it says GREEN as well as in the next code line ant change the number to change the thickness of the line.</p> <p>This code has a hardcoded confidence threshold for both the YOLO objectevness score and the class score, this can be found in the NMSBoxes method and the if confidence line accordingly. The main value to change first is the if confidence value.</p> <p>To close the image, you need to press the key ‘q’ as closing the display window is not going to work as it will reopen again.</p> <p>Note: This code allows the pop-up window, which displays the detections, to be closed only when the "q" key is pressed. Simply closing the window will not work.</p>
High spatiotemporal resolution free surface detection using cost-effective video equipment and computer vision techniques in nearly stationary flow along a transparent wall in the laboratory
<p>The identification of the air-water interface in free surface flows traditionally involves intrusive techniques or costly equipment. Non-intrusive alternatives, such as computer vision, are emerging as highly effective substitutes or supplements for more invasive techniques in laboratory measurements, thanks to their straightforward implementation and cost efficiency. This research specifically delves in the conjunction of various naive techniques, exploring their collective precision in detecting the air-water interface along transparent walls in laboratory. A detection technique based on the double gradient of the image is applied and thoroughly examined. The study progresses through multiple refinement stages, culminating in a method that is both cost effective and easy to implement. This methodology allows for large-scale, high resolution measurements (200 mm × 1800 frames per video at a 0.25 mm, 50 Hz resolution), offering both spatial and temporal measurements by adeptly detecting the free surface along transparent walls.</p>
EOL computer vision pipelines: Object Detection for Image Cropping: Aves
<p>Produced by an detection model pretrained on MS COCO 2017. Automatically crops images of birds (Aves) to square dimensions centered around animal(s). </p> <p>388,166 rows </p>
EOL computer vision pipelines: Object Detection for Image Cropping: Multi-taxon
<p>Produced by EOL Multitaxon Object Detection Model. Automatically crops images of snakes & lizards (Squamata), beetles (Coleoptera), frogs (Anura), and carnivores (Carnivora) to square dimensions centered around animal(s). Model available in the <a href="https://www.kaggle.com/models/eolorg/multitaxa-crops-thumbnails" target="_blank" rel="noopener">EOL Model Zoo on Kaggle</a>.</p> <ul> <li>Anura = 42646 rows</li> <li>Coleoptera = 115276 rows</li> <li>Squamata = 132680 rows</li> <li>Carnivora = 31132 rows</li> </ul>
EOL computer vision pipelines: Object Detection for Image Cropping: Chiroptera
<p>Produced by EOL Chiroptera Object Detection Model. Automatically crops images of bats (Chiroptera) to square dimensions centered around animal(s). Model available in the <a href="https://www.kaggle.com/models/eolorg/chiroptera-crops-thumbnails" target="_blank" rel="noopener">EOL Model Zoo on Kaggle</a>.</p> <p> </p> <p>17,401 rows</p>
EOL computer vision pipelines: Object Detection for Image Cropping: Lepidoptera
<p>Produced by EOL Lepidoptera Object Detection Model. Automatically crops images of butterflies and moths (Lepidoptera) to square dimensions centered around animal(s). Model available in the <a href="https://www.kaggle.com/models/eolorg/lepidoptera-crops-thumbnails" target="_blank" rel="noopener">EOL Model Zoo on Kaggle</a>.</p> <p> </p> <p>608,163 rows</p> <p> </p>
GTSRB - German Traffic Sign Recognition Benchmark by Real-Time Computer Vision at Ruhr-Universität Bochum
<div> <div>The German Traffic Sign Benchmark is a multi-class, single-image classification challenge held at the International Joint Conference on Neural Networks (IJCNN) 2011. <br>Our benchmark has the following properties: <br>- Single-image, multi-class classification problem <br>- More than 40 classes<br>- More than 50,000 images in total <br>- Large, lifelike database<br><br>Acknowledgements: [INI Benchmark Website][1]<br>[1]: http://benchmark.ini.rub.de/</div> </div>
REMODEL. WP4. Vision-Based Perception. T4-4. Functional component detection. Data related to a paper presented at 27th International Conference on Automation and Computing (ICAC) (2022)
<p>Dataset with evaluation parameters of the paper "Real-Time Instance Segmentation of Pedestrians using Transfer Learning", DOI: <a href="https://doi.org/10.1109/ICAC55051.2022.9911121">10.1109/ICAC55051.2022.9911121</a></p>
Dataset for "Computer vision assisted decomposition analysis of atom probe tomography data"
<p>Dataset for the article "Computer vision assisted decomposition analysis of atom probe tomography data". APT measurements were performed by Marcus Hans at Materials Chemistry (RWTH Aachen University) using a CAMECA LEAP 4000X HR. Training data was created by Janis A. Sälker.</p> <p>Content:</p> <p>- 13 (V,Al)N and 3 (Ti,Al)N APT reconstructions (.epos file format) and the corresponding range file (.rrng file format).</p> <p>- Training data (images & masks) for 9 labeled (V,Al)N APT samples (h5 file format). Image data with key "image" of shape (2, number_of_slices, 608, 192), where 2 corresponds to the V- and Al-contribution/channel and 608/192 to the height/width of the images. Masks/labels with key "label" of shape (number_of_slices, 608, 192)</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.