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.
328
datasets available to search
ShareScore release 0.9.0
Dataset results
328 results for “segmentation dataset”
Micro-CT Imaging Dataset on ex-vivo Ovine Functional Spinal Segments as Healthy, Injured and Treated with Cement Discoplasty
<p>General information:</p> <p>- This dataset contains micro-CT images and mechanical test data from ovine functional spinal units (FSU). <br> - The micro-CT data was produced using a Bruker SkyScan 1172. The settings for the scans are given in the '.log' files in each folder. <br> - The compression testing was conducted on an MTS 858 Mini Bionix T/II. The settings for each test can be found in test '.txt' files.<br> - In short, every FSU was mechanically tested in compression under different conditions. Before and after every test, the FSUs were scanned to ensure there was no damage<br> to the sample. More information can be found in the related publication: <br> - The mechanical testing data is arranged in folders with consecutive cycles. It is highly recommended to use the last three cycles for analysis. </p> <p>Data set notation:</p> <p>- All the datasets are noted by Sheep number. Sh7 = Sheep 7; Sh8 = Sheep 8; Sh9 = Sheep 9. In the publication, the numbers were switched to 1,2,3 respectively.<br> - files denoted with '_rec' contain the reconstruction of the projection images. <br> - 'Tested' or 'After test' files refers to the scan after mechanical testing. </p>
LumASe: A Lumbar Vertebra Anatomical Region Segmentation Dataset
<p>Lumbar vertebra anatomical region segmentation is crucial in an automated spine processing pipeline. To boost the research in automated lumbar vertebra anatomical region segmentation, we propose a new dataset called LumASe. The entire dataset consists of 663 vertebrae ranging from L1 to L5 which are cropped from lumbar spine CT scans. The data was acquired at ShengJing Hospital of China Medical University using three major manufacturers (Philips, Siemens and Toshiba). Cases with vertebral fractures, metallic implants, bone tumors and foreign materials are excluded. All 3D CT lumbar spine images have corresponding segmentation masks annotated at the voxel level by 3 physicians using the Pair annotation package. In each vertebra, we consider seven anatomical regions as the region of interest including superior articular process (SAP), vertebral body (VB), transverse process (TP), lamina (L) pedicle (P), spinous process (SP) and inferior articular process (IAP).</p>
Automated Segmentation of Large Image Datasets using Artificial Intelligence for Microstructure Characterisation and Damage Analysis
<p>Many properties of commonly used materials are driven by their microstructure, which can be influenced<br>by the composition and manufacturing processes. To optimise future materials, understanding the<br>microstructure is critically important. Here, we present two novel approaches based on artificial intelligence<br>that allow the segmentation of the phases of a microstructure for which simple numerical approaches, such<br>as thresholding, are not applicable: One is based on the nnU-Net neural network, and the other on generative<br>adversarial networks (GAN).<br>Using scanning electron microscopy images collected from large areas (~1 mm²) of dual-phase steels as a<br>case study, we demonstrate how both methods effectively segment intricate microstructural details,<br>including martensite, ferrite, and damage sites, for subsequent analysis.<br>Either method shows substantial generalizability across a range of image sizes and conditions, including<br>heat-treated microstructures with different phase configurations. The nnU-Net excels in mapping large<br>image areas. Conversely, the GAN-based method performs reliably on smaller images, providing greater<br>step-by-step control and flexibility over the segmentation process.<br>This study highlights the benefits of segmented microstructural data for various purposes, such as<br>calculating phase fractions, modelling material behaviour through finite element simulation, and<br>conducting geometrical analyses of damage sites and the local properties of their surrounding<br>microstructure.</p> <p>https://doi.org/10.1016/j.matdes.2024.113031</p>
Merged Multi-Organ Abdominal CT Segmentation Dataset
<p><strong>Version 2.0</strong></p> <p>This version maintains the same dataset as Version 1.0, the only addition is the CSV metadata file. Please note that the BTCV images do not include metadata information.<br></p> <p> </p> <p><strong>Description</strong></p> <p>The abdominal CT images and reference segmentations were drawn from three datasets: the <a href="https://doi.org/10.7303/syn3193805" target="_blank" rel="noopener">Beyond the Cranial Vault (BTCV) Abdomen dataset</a> [1], the <a href="https://amos22.grand-challenge.org/" target="_blank" rel="noopener">Multi-Modality Abdominal Multi-Organ Segmentation Challenge 2022 dataset</a> [2], and the <a href="../records/10047292">TotalSegmentator dataset</a> [3]. Given the class differences among the three datasets, their consolidation requires the elimination of several classes, resulting in a unified dataset of 680 CT images comprising 12 classes common to all three, including:</p> <ul> <li>Spleen</li> <li>Right Kidney</li> <li>Left Kidney</li> <li>Gallbladder</li> <li>Esophagus</li> <li>Liver</li> <li>Stomach</li> <li>Aorta</li> <li>Inferior Vena Cava</li> <li>Pancreas</li> <li>Right Adrenal Gland</li> <li>Left Adrenal Gland</li> </ul> <p>The original work for which this dataset was created can be found in <a href="https://github.com/nicoalpis/Robust-Abdominal-Segmentation" target="_blank" rel="noopener">this GitHub repository</a>. </p> <p> </p> <p><strong>Terms of use</strong></p> <p>The terms of use of this data set include the terms of use of the <a href="https://doi.org/10.7303/syn3193805">Beyond the Cranial Vault (BTCV) Abdomen dataset</a> (<a href="https://doi.org/10.7303/syn3193805">terms of use</a>; after <a href="https://www.synapse.org/#!Synapse:syn3193805/wiki/217753">registration</a>, you can <a href="https://www.synapse.org/#!Synapse:syn3376386">access the data</a>), <a href="https://amos22.grand-challenge.org/" target="_blank" rel="noopener">Multi-Modality Abdominal Multi-Organ Segmentation Challenge 2022 dataset</a> (<a href="https://doi.org/10.5281/zenodo.7155725" target="_blank" rel="noopener">terms of use and data access</a>), and <a href="../records/10047292">TotalSegmentator dataset</a> (<a href="../records/10047292">terms of use and data access</a>). If you use these reference segmentations, please cite the references below.</p> <p><strong>References</strong></p> <p>[1] Landman BA, Xu Z, Igelsias JE, Styner M, Langerak TR, and Klein A, "MICCAI multi-atlas labeling beyond the cranial vault - workshop and challenge," 2015, https://doi.org/10.7303/syn3193805.</p> <p>[2] Ji, Yuanfeng, et al. "Amos: A large-scale abdominal multi-organ benchmark for versatile medical image segmentation." Advances in Neural Information Processing Systems 35 (2022): 36722-36732.</p> <p>[3] Wasserthal, J., Breit, H. C., Meyer, M. T., Pradella, M., Hinck, D., Sauter, A. W., ... & Segeroth, M. (2023). Totalsegmentator: Robust segmentation of 104 anatomic structures in ct images. Radiology: Artificial Intelligence, 5(5).</p>
A dataset for robotic outdoor visual navigation with multiple passages through trajectory segments
<p>The images were captured by a fisheye camera and a magnetic compass was used to acquire the orientation data. The datasets are split in two folders:<br> 1) LEARN: In order to learn a new place, the robot camera captures 15 images over a 360 degrees panorama. During this process, the robot stays still in order to avoid distortions in the representation of the place.<br> 2) EXPLO: When exploring the environment (i.e. the rest of the time), the robot only captures 7 images per panorama, for the purpose of faster place recognition. Images are captured while the robot is moving. Various exploration panoramas are recorded around the trajectory performed in the learning panoramas (see traj.pdf).<br> <br> The average distance between two learning panoramas is 0.93 +/- 0.03 meters<br> The average distance traveled during an exploration panoramas is 0.71 +/- 0.01 meters<br> <br> DATASET A<br> ---------<br> - 20 meters long<br> - 22 learning panoramas (i.e. sets of 15 images captured while robot is stopped)<br> - 5 exploration trajectories<br> - A_on_learned: 29 exploration panoramas (i.e. sets of 7 images captured while robot is moving)<br> - A_parallel: 29 exploration panoramas<br> - A_diagonal1: 28 exploration panoramas<br> - A_diagonal2: 30 exploration panoramas<br> - A_diagonal3: 29 exploration panoramas<br> <br> DATASET B<br> ---------<br> - 20 meters long<br> - 21 learning panoramas (i.e. sets of 15 images captured while robot is stopped)<br> - 4 exploration trajectories<br> - B_on_learned: 29 exploration panoramas (i.e. sets of 7 images captured while robot is moving)<br> - B_parallel: 29 exploration panoramas<br> - B_diagonal1: 29 exploration panoramas<br> - B_diagonal2: 29 exploration panoramas<br> <br> DATASET C<br> ---------<br> - 23.1 meters long<br> - 25 learning panoramas (i.e. sets of 15 images captured while robot is stopped)<br> - 2 exploration trajectories<br> - C_on_learned: 34 exploration panoramas (i.e. sets of 7 images captured while robot is moving)<br> - C_parallel: 34 exploration panoramas<br> <br> <br> <br> PANO_INFO FILE STRUCTURE<br> ------------------------<br> Every folder containing images also contains an info file, named either learn_pano_info.SAVE or explo_pano_info.SAVE. Each line corresponds to an image. The structures is the following:<br> - column 1: id = image_id + 1<br> - column 2: azimuth of the center of the image in degrees/360 (value in [0,1])<br> - column 3: elevation of the center of the image. irrelevant in this database (equal to 0).<br> - column 4: type of panorama: equal to 1 if learning and to 0 if exploration.<br> - column 5: end of panorama: equal to 1 if it corresponds to the last image of a panorama.<br> <br> <br> REFERENCES<br> ----------<br> The dataset was used in the paper: Belkaid, M., Cuperlier, N., and Gaussier, P. Combining local and global visual information in context-based neurorobotic navigation. In Proceedings of the IEEE International Joint Conference on Neural Networks (IJCNN), pages 4947-4954, doi: 10.1109/IJCNN.2016.7727851, 2016.<br> <br> </p>
Dataset of Axon Segmentation and Centerlines, Acquired using a Two-photon Microscope in the Live Mouse Cortex
<p>This dataset contains 20 images of real axons that were published previously in Bass et al (2017). This subset dataset has been added labels of binary images for segmentation of the axons, and of centerline for tracing of the axons.</p> <p>We provide the following:</p> <ul> <li>Images-MAX: 2D images of axons (.png)</li> <li>Images-TIFF: 3D images of axons (.tiff)</li> <li>Labels-binary: Manual segmentation of the axons as binary images (.png)</li> <li>Labels-tracing: Manual tracing of the axons (.swc)</li> </ul> <p>This data was collected in the live mouse cortex, using a two-photon microscope, with a 40x objective, at zoom 4, and with a resolution of 512 × 512 pixels, 0.147 <em>μ</em>m per pixel for the <em>x</em>, <em>y</em> planes, and 1 <em>μ</em>m for the <em>z</em> plane. . </p> <p><strong>Please cite the following papers when using this dataset:</strong></p> <p>T. Dai, M. Dubois, K. Arulkumaran, J. Campbell, B. Billot, C. Bass, Z. Uslu, V. De Paola, C. Clopath, and A. A. Bharath. Deep reinforcement learning for subpixel neural tracking. <em>Medical Imaging with Deep Learning</em>. 2019.</p> <p>Bass C, Helkkula P, De Paola V, Clopath C, Bharath AA. Detection of axonal synapses in 3D two-photon images. Giniger E, ed. <em>PLoS ONE</em>. 2017;12(9):e0183309. doi:10.1371/journal.pone.0183309.</p>
Dataset: A scalable method to improve gray matter segmentation at ultra high field MRI.
<p><strong>Dataset description: </strong>Accompanying data for manuscript “<a href="https://www.biorxiv.org/content/early/2018/01/10/245738">A scalable method to improve gray matter segmentation at ultra high field MRI</a>” written by Omer Faruk Gulban, Marian Schneider, Ingo Marquardt, Roy Haast, Federico De Martino.</p> <p><a href="http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0198335">Published in PLOS One, June 6, 2018</a>.</p> <p>The dataset consist of 7 Tesla MRI anatomical images of living human brains (whole brain; 0.7mm isotropic resolution; T1 weighted, T2* weighted, proton density weighted MPRAGE images; inversion 1, inversion 2, T1, uni, MP2RAGE images; Multi-echo 3D GRE) and hand labeled cortical gray matter images (for further details see <a href="http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0198335#sec012">section 4.1 of our manuscript</a>).</p> <p>Folder structure is organized according to Brain Imaging Data Structure (BIDS). Further details can be found the README files.</p> <p><strong>Citation</strong></p> <p>Please cite the following paper together with this dataset doi:</p> <ul> <li>Gulban, O. F., Schneider, M., Marquardt, I., Haast, R. A. M., & De Martino, F. (2018). A scalable method to improve gray matter segmentation at ultra high field MRI. <em>PLOS ONE</em>, <em>13</em>(6), e0198335. http://doi.org/10.1371/journal.pone.0198335</li> </ul> <p><br> Bibtex format:</p> <p>```<br> @article{Gulban2018,<br> author = {Gulban, Omer Faruk and Schneider, Marian and Marquardt, Ingo and Haast, Roy A. M. and {De Martino}, Federico},<br> doi = {10.1371/journal.pone.0198335},<br> editor = {Pham, Dzung},<br> issn = {1932-6203},<br> journal = {PLOS ONE},<br> month = {jun},<br> number = {6},<br> pages = {e0198335},<br> title = {{A scalable method to improve gray matter segmentation at ultra high field MRI}},<br> url = {http://dx.plos.org/10.1371/journal.pone.0198335},<br> volume = {13},<br> year = {2018}<br> }<br> <br> ```</p>
Annotated dataset for sub-shot segmentation evaluation
<p>This dataset was created for evaluating the performance of the developed motion-driven approach for fine-grained temporal segmentation of user-generated videos, that is reported in [1].</p> <p>Given the fact that user-generated videos are most commonly captured without interruption using a single camera, thus being single-shot videos, their fine-grained temporal segmentation aims to identify visually coherent parts (called sub-shots) that correspond to individual actions taking place during the video recording, such as left camera panning, camera zoom-out or tracking of a moving object.</p> <p>Based on the above, the created dataset contains the ground-truth sub-shot segmentation for 33 single-shot videos. This ground-truth segmentation was created by human annotation of the sub-shot boundaries for each video, where each boundary indicates the end of a visually and temporally contiguous activity of the video recording device and the start of the next one (e.g. a downward camera tilting that is followed by a camera zoom-in). Overall, our dataset contains 674 sub-shot transitions.</p> <p>The set of videos can be divided in three parts:</p> <ol> <li><em>Own Videos</em>: 15 single-shot videos of total duration 6 minutes, which contain clearly defined fragments that correspond to several video recording activities;</li> <li><em>Amateur Videos</em>: 5 single-shot amateur videos of total duration 17 minutes, found on the YouTube platform;</li> <li><em>Movie Excerpts</em>: 13 single-shot parts of known movies of total duration 46 minutes which represent professional video content.</li> </ol> <p>The videos of the first part were recorded in the external spaces of our (CERTH-ITI) facilities using an iPhone 5 smart-phone. These videos and their ground-truth are freely provided through the "Provided files" section below.</p> <p>The videos of the second and third part can be found on the YouTube platform (links in the "Video collection" section below). For these videos we only provide the ground-truth data again through the "Provided files" section below.</p> <p>For each video of the dataset we provide:</p> <ul> <li>The video ID;</li> <li>The video filename for "Own videos" and the YouTube video name and URL for "Amateur videos" and "Movie Excerpts" (accessed on 12/11/2017);</li> <li>The video frame-rate that directly affects the ground-truth annotation; the videos of the second and third part should be downloaded at the corresponding frame-rates (default values when using the KeepVid web application - <a href="https://keepvid.com/">https://keepvid.com/</a>) to match the created annotations.</li> </ul> <p>The above information is included in the provided "video_collection_info.pdf" file.</p> <p>The needed files for using the created dataset are in the provided compressed "dataset.zip" file. After unpacking the compressed file a structure of directories will be generated. In this structure:</p> <ol> <li>The "videos/own_videos" directory contains: a) the video files of the first part of the dataset, and b) the "list.txt" file which lists the video id, the filename and the frame-rate for each video of the "Own videos" collection, in a tab-separated format.</li> <li>The "videos/amateur_videos" directory contains the "list.txt" file which lists the video id, the filename and the frame-rate for each video of the "Amateur videos" collection, in a tab-separated format.</li> <li>The "videos/movie_excerpts" directory contains the "list.txt" file which lists the video id, the filename and the frame-rate for each video of the "Movie excerpts" collection, in a tab-separated format.</li> <li>The "ground_truth" directory contains 33 txt files (one per video) where each file is named after the ID of the corresponding video and includes the annotation of the sub-shots in the video; each line of this file corresponds to the starting frame of a sub-shot of the video (using a zero-based index of frames).</li> <li>The "evaluation" directory contains a simple Matlab script, called "eval_segm.m", that was used for evaluating the results of the sub-shot segmentation analysis.</li> <li>A readme file that contains the documentation of the dataset (i.e. the information that is available in this webpage).</li> </ol> <p>CERTH-ITI holds the copyright of the “<em>Own videos</em>” part of the dataset. These videos can be freely used for academic/research purposes only and their public playback is not allowed. CERTH-ITI does not own the copyright of the “<em>Amateur Videos</em>” and “<em>Movie Excerpts</em>” parts of the dataset. The use of these videos must abide by the YouTube copyright policy (<a href="https://www.youtube.com/yt/about/copyright/">https://www.youtube.com/yt/about/copyright/</a>). Any users of these videos must accept full responsibility for the use of these parts of the dataset, including but not limited to the use of any copies of copyrighted videos that they may create from the dataset.</p> <p>If you use this dataset, please cite the following scientific work:</p> <p>[1] K. Apostolidis, E. Apostolidis, V. Mezaris, <em>"A motion-driven approach for fine-grained temporal segmentation of user-generated videos"</em>, Proc. 24th Int. Conf. on Multimedia Modeling (MMM2018), Bangkok, Thailand, Feb. 2018.</p> <p> </p> <p> </p>
Pecha Line Segmentation Datasets
<p>Training data for line segmentation of Tibetan woodblock prints. This version of dataset contain images randomly picked images from LOC scans of the <a href="https://www.tbrc.org/#!rid=W4CZ5369">derge kangyur</a>.</p> <p>This dataset was trained with Unet Model using fastai library. The result was unsatisfactory due to insufficient data.</p> <p><strong>Datasets contains:</strong></p> <ol> <li>Images/ : Original pecha image</li> <li>Labels/ : Mask of each image</li> <li>valid.txt: list of validation filename</li> <li>code.txt: list of name of object</li> </ol>
CHAS - Cultural Heritage Architectural Segmentation dataset
<p>CHAS is a point cloud dataset from cultural heritage aimed to provide data for semantic segmentation techniques. The data in this repository were generated by terrestrial laser scanning and pictures from aerial survey using UAV. This dataset comprises relevant buildings representing religious and colonial Brazilian architecture. We hope to allow future research in point cloud segmentation and automatic Building Information Modeling.</p> <p>We use the .7z file format to compact the clouds, providing high compression ratio. File names with "raw" suffix refers to RGB registered cloud. The "gt" suffix stands for ground truth clouds.</p> <p> </p> <p><strong>Raw point cloud description</strong></p> <p>Name Architectural Style Construction Year TLS UAV Points* ID</p> <p>Good Death's Church Neoclassical 1867 Yes Yes 210 <em>boa_morte</em><br> Central Mill Industrial 1881 Yes Yes 26 <em>engenho</em><br> Ball House Modern 1940 Yes No 370 <em>baile</em><br> Quilombo Farm Colonial 1891 Yes Yes 55 <em>quilombo</em> St. Francis of Assisi Church Modern 1943 Yes Yes 46 <em>curia</em></p> <p>* millions </p> <p> </p> <p><strong>Equipment</strong></p> <ul> <li>FARO Focus3D X 330 HDR*</li> <li>DJI Inspire 2 with Gimbal DJI Zenmuse X4S 4K | H264 | F2.8 - F11 | 20MP**</li> <li>DJI Spark 12 MP Full HD GPS **</li> </ul> <p><em>*Terrestrial Laser Scanner (TLS)</em></p> <p><em>**Unmanned Aerial Vehicle (UAV)</em></p> <p> </p> <p><strong>Software</strong></p> <p>The following third-party software have been used to register the clouds.: FARO SCENE, CloudCompare and Pix4D Mapper. For a better understanding of the hybrid point cloud registration approach used in this dataset we recommend the following article: <a href="http://papers.cumincad.org/cgi-bin/works/paper/ecaade2018_295">BIM for Heritage Documentation An ontology-based approach</a>.</p> <p> </p> <p><strong>Ground-truth</strong></p> <p>Ground-Truth in this dataset was manually created with the Interactive Segmentation Tool on CloudCompare. Distinct colors represent different architectural elements. An “optimal” segmentation algorithm must partition the raw point clouds into equivalent single-color ground-truth clusters. </p> <p> </p> <p><strong>How to cite this dataset?</strong></p> <p>Paiva, P. V. V., Cogima, C. K., Dezen-Kempter, E. and Carvalho, M. A. G. "Historical building point cloud segmentation combining hierarchical watershed transform and curvature analysis." <em>Pattern Recognition Letters</em> 135 (2020): 114-121. DOI: <a href="https://doi.org/10.1016/j.patrec.2020.04.010">https://doi.org/10.1016/j.patrec.2020.04.010</a></p> <p>Paiva, P. V. V., Cogima, C. K., Dezen-Kempter, E. and Carvalho, M. A. G. "<em>CHAS - Cultural Heritage Architectural Segmentation dataset</em>". Zenodo, March 26 2019. DOI: <a href="https://doi.org/10.5281/zenodo.2609498">https://doi.org/10.5281/zenodo.2609498</a></p> <p> </p>
Fiber and vessel dataset for segmentation and characterization
<p>This repository hosts a comprehensive collection of datasets used to develop an innovative deep learning model designed to enhance the segmentation and characterization of macerated fibers and vessel forms in microscopy images. Included in the deposit are raw images, alongside meticulously prepared training and validation datasets. We present an automated segmentation approach that utilizes the one-stage YOLOv8 model, which has been specifically adapted to process high-resolution microscopy images up to 32640 x 25920 pixels. Our model excels in cell detection and segmentation, demonstrating exceptional proficiency.</p>
A semantic segmentation dataset of Arctic sea ice from Operation IceBridge data
<p>This dataset is a semantic segmentation dataset of Arctic sea ice based on deep learning method from Operation IceBridge images. It contains 29,372 labeled images, each of which corresponds to an image of Operation IceBridge and are stored in TIFF format. The dataset can be accessed using ArcGIS, ENVI, and the GDAL library in Python easily. Where label 1 represents melt ponds, label 2 represents sea ice/snow, label 3 represents submerged ice, and label 4 represents open ocean water, respectively.</p>
LiDAR Segmentation Benchmark Dataset: La Palma
<p>The purpose of this benchmark dataset is primarily to allow the assessment of individual tree segmentation algorithms on LiDAR data. The high density ULS point cloud data were captured by the Zenmuse L1 sensor on a DJI Matrice 300RTK over a coniferous plot on La Palma (Canary Islands, Spain). The dataset comprises the CHM and the normalized 3D cloud, as well as the fully annotated data to serve as a benchmark for segmentation.</p> <p>The publication associated with this database is:</p> <p>Marcello, J.; Spínola, M.; Albors, L.; Marqués, F.; Rodríguez-Esparragón, D.; Eugenio, F. Performance of Individual Tree Segmentation Algorithms in Forest Ecosystems Using UAV LiDAR Data. <em>Drones</em> <strong>2024</strong>, <em>8</em>, 772. https://doi.org/10.3390/drones8120772</p>
Multi-resolution dataset for photovoltaic panel segmentation from satellite and aerial imagery
<p>A <a href="https://www.sciencedirect.com/topics/engineering/photovoltaics">photovoltaic</a> (PV) dataset from satellite and aerial imagery. The dataset includes three groups of PV samples collected at the spatial resolution of 0.8m, 0.3m and 0.1m, namely PV08 from Gaofen-2 and Beijing-2 imagery, PV03 from aerial photography, and PV01 from UAV orthophotos. PV08 contains rooftop and ground PV samples. Ground samples in PV03 are divided into five categories according to their background land use type: shrub land, grassland, cropland, saline-alkali, and water surface. Rooftop samples in PV01 are divided into three categories according to their background roof type: flat concrete, steel tile, and brick. Data document can refer to the preprint https://essd.copernicus.org/preprints/essd-2021-270/</p>
Dataset for semantic segmentation in NDT with step-heating thermography for CFRP laminates
<p>Dataset composed of 36 images (640x480 pixels) of 30 bands or channels from step-heating for the same Carbon Fiber Reinforced Polymer (CFRP) laminate.</p> <p>In each image the specimen is rotated 10º to generate new data, with different illumination, background, and heating/cooling sequences.</p> <p>Images are generated from a video composed of the heating process, which takes 10 seconds, and the cooling process, which takes another 10 seconds, to a total of 1000 frames per video. This video is then reduced to 30 images using different processing methods explained below.</p> <p>The 30 bands or channels consist of the following bands repeated for the heating and cooling processes:</p> <ol> <li>First PCT component</li> <li>Second PCT component</li> <li>Third PCT component</li> <li>Fourth PCT component</li> <li>PPT</li> <li>Kurtosis</li> <li>Skewness</li> <li>TSR Coefficient 7</li> <li>TSR Coefficient 6</li> <li>TSR Coefficient 5</li> <li>TSR Coefficient 4</li> <li>TSR Coefficient 3</li> <li>TSR Coefficient 2</li> <li>TSR Coefficient 1</li> <li>TSR Coefficient 0</li> </ol> <p>Please cite the original paper:</p> <p>LINK: TODO</p> <p>BibTex:</p> <p>TODO</p> <p> </p> <p> </p>
Grapevine Dataset for plant organ segmentation
<p><br>This is a dataset for Grapevine segmentation for the purposes of winter pruning, created in a joint work between Istituto Italiano di Tecnologia and Università Cattolica del Sacro Cuore, as part of the Vinum project. The work was also cofounded by Italian Ministry of University and Research PRIN 20172HHNK5 Project, under the supervision of Matteo Gatti, Claudio Semini and Darwin Caldwell.</p> <p>The images were captured in the simulated grapevine garden in Università Cattolica, located in Piacenza, Italy, and it is split in two parts, following agronomic trials being run on the university. The first part is a group of seven specimen that is the Control group, and the second group of eight specimens is being performed shoot thinning.</p> <p>Each plant has around 5 spurs, and the photos are taken from both sides of the grapevine specimen, meaning that there is one picture where the plant grows from left to right, and the other side where the plant grows from right to left. The resolution of the images is 4608x3456, and we have a total of 149 images. </p> <p> </p> <p>It is annotated using the COCO Segmentation format, and some of the statistics of the dataset such as the number of annotations and number of images can be seen next.</p> <p><strong>Control Complex</strong></p> <ul> <li>Total <ul> <li>Images: 69</li> <li>Annotated Images: 69</li> <li>Annotations: 1838</li> <li>Categories: 5</li> </ul> </li> <li>Annotations Per Category <ul> <li>Main Cordon: 79</li> <li>Cane: 440</li> <li>Node: 1100</li> <li>Arm: 103</li> <li>Spur: 116</li> </ul> </li> <li>Annotated Images Per Category <ul> <li>Main Cordon: 69</li> <li>Cane: 69</li> <li>Node: 69</li> <li>Arm: 62</li> <li>Spur: 68</li> </ul> </li> </ul> <p><strong>Shoot Thining Simple</strong></p> <ul> <li>Total <ul> <li>Images: 79</li> <li>Annotated Images: 79</li> <li>Annotations: 1635</li> <li>Categories: 5</li> </ul> </li> <li>Annotations Per Category <ul> <li>Main Cordon: 84</li> <li>Cane: 341</li> <li>Node: 912</li> <li>Arm: 154</li> <li>Spur: 144</li> </ul> </li> <li>Annotated Images Per Category <ul> <li>Main Cordon: 79</li> <li>Cane: 79</li> <li>Node: 79</li> <li>Arm: 79</li> <li>Spur: 77</li> </ul> </li> </ul> <p> </p>
Dataset for "Root Length Estimation: Automated Minirhizotron Image Analysis with Convolutional Networks without Segmentation"
<p>This data contains 4015 root images, splitted into 4 datasets, acquired using two minirhizotron (MR) system types - manual (Dataset 1 & Dataset 4) and automated (Dataset 2 & Dataset 3). It includes four crop species (corn, pepper, melon, and tomato) grown under various abiotic stresses. The data was acquired by researchers from Ben-Gurion University of the Negev, Beer Sheva, Israel, and used for research of automated TRL estimation with Convolutional Neural Networks.</p> <p>The annotations were conducted manually using the Rootfly software (Wells and Birchfield, Clemson University, South Carolina, USA), and data were transformed as CSV formats. In this software, the annotator must draw a root by marking points along the selected root. These points usually correspond to the coordinates at the start and the end of the root, and curving points along the root. These points are then connected in a line, the length of which reflects the real length of the selected root. The annotations has been done for all roots within an image, and for all images in the provided dataset.</p> <p>The provided annotations include the total root length (TRL) per image (mm) and the coordinates of annotated points.</p> <p>The annotations are given in two types of files:</p> <p>"TRL.csv" files: contain the image names and corresponding TRL values (mm).</p> <p>"pointsOutput.csv" files: contain the annotated image names and the coordinates of the points of the roots in the image (if the image contains roots) in the form of x1, y1, x2, y2, x3, y3, etc. It the image doesn't have roots, the file contains only its name.</p>
Wikimedia wikis' wiki segmentation dataset
<p>We maintain a <strong>wiki comparison dataset</strong> (which we used to call a wiki <em>segmentation</em> dataset) to show a simple, snapshot comparison of our wikis (as opposed to, say, Wikistats 2, which is mean to show simple trends within individual wikis or wiki groups).</p>
VegAnn: Vegetation Annotation of a large multi-crop RGB Dataset acquired under diverse conditions for image segmentation
<p> VegAnn - Vegetation Annotation - dataset, a collection of 3795 multi-crop RGB images acquired for different phenological stages using different systems and platforms in diverse illumination conditions. </p>
WE3DS: An RGB-D image dataset for semantic segmentation in agriculture
<p>Here, we introduce a novel RGB-D image database (WE3DS) for semantic segmentation in crop farming. It contains 2,568 RGB-D images (color image and distance map) and hand-annotated ground-truth masks for semantic segmentation and is the first RGB-D image dataset for multi-class plant species semantic segmentation task. Images were taken under natural light conditions using an RGB-D sensor consisting of two RGB cameras in a stereo setup.</p> <p> </p> <p><strong>Please cite the original source when using this dataset.</strong></p> <p>Kitzler, F.; Barta, N.; Neugschwandtner, R.W.; Gronauer, A.; Motsch, V. WE3DS: An RGB-D Image Dataset for Semantic Segmentation in Agriculture. <em>Sensors</em> <strong>2023</strong>, <em>23</em>, 2713. <a href="https://doi.org/10.3390/s23052713">https://doi.org/10.3390/s23052713 </a></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.