Find research datasets worth reusing

opencc-by-4.0Jan 2020View details →

Synthetic images of cell nuclei in widefield microscopy

The images were generated by <a href="http://www.cs.tut.fi/sgn/csb/simcep/tool.html">SIMCEP</a>, a widefield fluorescence microscopy biological images simulator. The dataset is used to demonstrate the execution of image analysis workflows with BIAFLOWS on a local machine from a jupyter notebook.

opencc-by-4.0Dec 2023View details →

VoroCrack3d: An annotated data set of 3d CT concrete images with synthetic crack structures

VoroCrack3d is an annotated data set of 3d CT images of concrete with synthetic crack structures. Its main purpose is the training and testing of machine learning models for 3d crack segmentation. The data set comprises 1344 images together with their corresponding ground truths. The concrete backgrounds are cropped out sections of size 400x400x400 voxels of CT images of concrete. To this end, several different concrete samples were scanned (normal concrete (NC), high-performance concrete (HPC), ultra-high-performance concrete (UHPC), air pore concrete; without and with reinforcements (straight steel fibers, crimped steel fibers, hooked-end steel fibers, polypropylene fibers, fibers made of glass fiber-reinforced polymer). The original concrete images have a resolution between 2.8 and 106 micrometers. The crack structures are modeled via minimum-weight surfaces in Voronoi diagrams according to the paper [1] C. Jung, C. Redenbach, Crack Modeling via Minimum-Weight Surfaces in 3d Voronoi Diagrams, Journal of Mathematics in Industry, 13, 10 (2023). https://doi.org/10.1186/s13362-023-00138-1. The surfaces are discretized, dilated and superimposed on the concrete backgrounds. The data set offers a high variety regarding concrete types, noise levels and crack widths, shapes, regularity and branching. This makes it suitable for studying the generalizability and robustness of 3d crack segmentation methods. ______________________________________________________________________________________________ The folder 'data' contains seven subfolders, each containing the data generated from a specific concrete type (NC, HPC, air pore concrete, polypropylene fiber-reinforced concrete, steel fiber-reinforced concrete (straight, crimped and hooked-end steel fibers)). Each subfolder again contains four subfolders according to the point process model that was used for generating the 3d Voronoi diagrams. The point processes and Voronoi diagrams are restricted to windows of size 400x150x400.  - 'hc': Hard core point process with 60% volume density and intensity 0.000025 obtained from force-biased sphere packing. - 'matclust': Matérn cluster process with parent intensity 0.0002/50, offspring intensity 50 and cluster radius 20. - 'ppp': Poisson point process with intensity 0.0002. - 'ppp-scaled': Poisson point process with intensity 0.0002 (but inside 200x150x200 window). The resulting Voronoi diagram is stretched in x- and z- direction by a factor of 2. Each of these contains five subfolders: one for the 3d input images, two for the corresponding labels (ground truths; one with and one without pores/fibers), one for the input and label previews (slice z=200 for each of the images) and a misc folder containing the concrete background without crack and, if applicable, the pore/fiber segmentation image. The data itself then contains 48 images: 1a-1d: crack with up to seven branches; fixed crack width (~1 voxel). 2a-2d: crack with up to four branches; fixed crack width (~1 voxel). 3a-3d: crack with up to one branch; fixed crack width (~1 voxel). 4a-4d: crack with no branches; fixed crack width (~1 voxel). 5a-5d: crack with no branches; fixed crack width (~3 voxels). 6a-6d: crack with no branches; fixed crack width (~5 voxels). 7a-7d: crack with no branches; fixed crack width (~7 voxels). 8a-8d: crack with up to seven branches; multiscale crack (bernoulli parameter 0.01); 9a-9d: crack with up to seven branches; multiscale crack (bernoulli parameter 0.02); 10a-10d: crack with up to seven branches; multiscale crack (bernoulli parameter 0.05); 11a-11d: crack with up to seven branches; multiscale crack (bernoulli parameter 0.1); 12a-12d: crack with up to seven branches; multiscale crack (bernoulli parameter 0.2); The names 'a'-'d' indicate level of added noise added to the image: a: None. b: Uniformly on [-sigma,sigma]  c: Uniformly on [-2*sigma,2*sigma]  d: Uniformly on [-4*sigma,4*sigma]  Negative values are mapped to 0.  For inputs of type int, noise values are rounded to the nearest integer. (sigma = standard deviation of voxel greyvalues in image) Note that the grey values in the ground truths correspond to the local crack width. They can be thresholded to obtain binary masks. For more details, we refer to [1].

opencc-by-nc-sa-4.0Jan 2024View details →

High-Voltage Disconnector State Identification: synthetic and real images of substation disconnectors

This dataset contains the training and test images used in the work detailed in the article: Barpp Gomes, V., Marchesi, B., Gruber, Y.A. et al. Exploring Synthetic Data for Training Deep Learning Models for High-Voltage Disconnector State Identification. J Control Autom Electr Syst (2025). <a href="https://doi.org/10.1007/s40313-025-01204-2">https://doi.org/10.1007/s40313-025-01204-2</a> Contais about 940,000 synthetic (CGI-rendered) and 60,000 real (camera-captured) samples of four types of substation disconnectors, on both open and closed states: <ul> <li>230 kV center break (type 1, as indicated in the article);</li> <li>230 kV center break (type 2);</li> <li>230 kV double side break</li> <li>525 kV horizontal semi-pantograph</li> </ul> Each zip file contains images of one type of substation disconnector. Images are sized 320x128 and are organized in folders, as follows: <ul> <li>00_train_synth: Synthetic training images.</li> <li>01_train_real: A small set of real training images, as indicated in the article.</li> <li>02_test_real_normal1: One set of real test images.</li> <li>03_test_real_normal2: Another set of real test images, from a different time period.</li> <li>04_test_real_maneuvers: A special set of real test images in which the switches have been operated (are in different states).</li> </ul>

opencc-by-4.0Aug 2019View details →

Synthetic dataset accompanying Neural Image Compression for Gigapixel Histopathology Image Analysis

This dataset was used to develop and evaluate the main method proposed in the paper "Neural Image Compression for Gigapixel Histopathology Image Analysis" published in IEEE Transactions on Pattern Analysis and Machine Intelligence with DOI 10.1109/TPAMI.2019.2936841. Please refer to the paper for a detailed description of the dataset. The dataset consists of a set of 50000 images and 50000 associated ground truth masks, distributed into training and test partitions. The name of each file follows the pattern "{id}_{tilted_label}_{nontilted_label}_{tilted_size}_{nontilted_size}_{kind}.png" where:   * id: unique identifier within each partition.   * tilted_label: image-level label corresponding to the tilted rectangle.   * nontilted_label: image-level label corresponding to the non-tilted rectangle.   * tilted_size: longest size of the tilted rectangle.   * nontilted_size: longest size of the non-tilted rectangle.   * kind: either "tile" or "mask" image type. The images are distributed into several data partitions used during cross-validation and fully described in "mnist_folds_set.json". Please rename "mnist_folds_set.json.removethis" into "mnist_folds_set.json". The code to recreate this dataset can be found in https://github.com/davidtellez/neural-image-compression.

opencc-by-4.0May 2023View details →

Synthetic Particle Image Dataset (SPID)

SPID is a comprehensive dataset composed of synthetic particle image velocimetry (PIV) image pairs and their corresponding exact optical flow computations. It serves as a valuable resource for researchers and practitioners in the field. The dataset is organized into three subsets: training, validation, and test, distributed in a ratio of 70%, 15%, and 15%, respectively.Each subset within SPID consists of an input denoted as "x", which comprises synthetic image pairs. These image pairs provide the necessary context for the optical flow computations. Additionally, an output termed "y" is provided, which represents the exact optical flow calculated for each image pair. Notably, the images within the dataset are single-channel, and the optical flow is decomposed into its u and v components.The shape of the input subsets in SPID is given by (number of samples, number of frames, image width, image height, number of channels), representing the dimensions of the input data. On the other hand, the shape of the output subsets is given by (number of samples, velocity components, image width, image height), denoting the shape of the optical flow data.It is important to mention that SPID dataset is a preprocessed version of the Raw Synthetic Particle Image Dataset (RSPID), ensuring improved usability and reliability. Moreover, the dataset is packaged as a NumPy compressed NPZ file, which conveniently stores the inputs and outputs as separate NumPy NPZ files with the labels train, validation and test as acess keys. This format simplifies data extraction and integration into machine learning frameworks and libraries, facilitating seamless usage of the dataset.SPID incorporates various factors that impact PIV analysis to provide a comprehensive and realistic simulation. The dataset includes image pairs with an image width of 665 pixels and an image height of 630 pixels, ensuring a high level of detail and accuracy with an 8-bit depth. It incorporates different particle radii (1, 2, 3, and 4 pixels) and particle densities (15, 17, 20, 23, 25, and 32 particles) to capture diverse particle configurations.To simulate real-world scenarios, SPID introduces displacement variations through the delta x factor, ranging from 0.05% to 0.25%. Noise levels (1, 5, 10, and 15) are also incorporated to mimic practical PIV measurements with varying degrees of noise. Furthermore, out-of-plane motion effects are considered with standard deviations of 0.01, 0.025, and 0.05 to assess their impact on optical flow accuracy.The dataset covers a wide range of flow patterns encountered in fluid dynamics. It includes Rankine uniform, Rankine vortex, parabolic, stagnation, shear, and decaying vortex flows, allowing for comprehensive testing and evaluation of PIV algorithms across different scenarios.By leveraging the SPID dataset, researchers can develop and validate PIV algorithms and techniques under various challenging conditions. Its realistic and diverse simulation of particle image velocimetry scenarios makes it an invaluable tool for advancing the field and improving the accuracy and reliability of optical flow computations.

opencc-by-4.0Apr 2020View details →

Synthetic spherical tissue images

A time-lapse sequence of synthetic cell membrane images along with their segmentations. Cell-lineages associating cell labels from consecutive time points are also provided.   File information: Image files are in the .inr.gz format, which can be read for instance using the timagetk (<a href="https://gitlab.inria.fr/mosaic/timagetk">https://gitlab.inria.fr/mosaic/timagetk</a>) Python library.

opencc-by-4.0Mar 2022View details →

Zellige example dataset: synthetic image dataset

Phantom 3D image containing three distinct and superimposed synthetic surfaces. It models a typical stack of confocal images of epithelial and non-epithelial structures.The surfaces generated are of two types: “solid” surfaces, presenting a homogeneous signal over the entire surface, or surfaces presenting a signal restricted to a polygonal mesh mimicking the mesh of apical cellular junctions of an epithelium observed at its surface. This dataset contains both the ground-truth height maps and the height maps generated with Zellige. The Zellige parameters used are: \(T_{A}=16, T_{otsu}=12, S_{min}=5, \sigma_{xy}=4, \sigma_{z}=2, T_{OSE1}=0.9, R_{1}=5, C_{1}=0.1, T_{OSE2}=0.1, R_{2}=10, C_{2}=0.8.\) Nota: the ground-truth height maps can be directly compared to Zellige height maps by subtraction. See the accompanying paper: Extracting multiple surfaces from 3D microscopy images in complex biological tissues with the Zellige software tool. Trébeau et al. 2022: <a href="https://doi.org/10.1101/2022.04.05.485876">https://doi.org/10.1101/2022.04.05.485876</a>

opencc-by-4.0Sep 2022View details →

Synthetic cryo electron microscopy single particle images containing biomolecular complexes with continuous conformational variability used for validating DeepHEMNMA method and validation results

This archive contains a synthetic dataset used for validating DeepHEMNMA method and the validation results. DeepHEMNMA is a deep learning extension of HEMNMA approach for analyzing continuous conformational variability of biomolecular complexes in cryo electron (cryo-EM) microscopy single particle images. We provide a training set of 20,000 images and an inference set of 50,000 images. The training images were used (1) to estimate the conformational and rigid-body parameters with HEMNMA and (2) to train the neural network using the parameters previously estimated with HEMNMA (the file with the HEMNMA-estimated parameters is provided). The inference images were used to infer the parameters with the trained neural network. Also, we provide (1) the input PDB structure, its normal modes, and the conformational and rigid-body parameters used to synthesize the 20,000 training images (ground-truth parameters) and (2) the conformational and rigid-body parameters inferred from the set of 50,000 inference images. The DeepHEMNMA method and the method for synthesizing images have been fully described in the following article: "Hamitouche I and Jonic S (2022), DeepHEMNMA: ResNet-based hybrid analysis of continuous conformational heterogeneity in cryo-EM single particle images. Front Mol Biosci 9, 965645. <a href="https://doi.org/10.3389/fmolb.2022.965645">https://doi.org/10.3389/fmolb.2022.965645</a> (in press)". Additionally, this article describes a test of DeepHEMNMA using one experimental cryo-EM dataset (available in EMPIAR database under the accession code EMPIAR-10016).

opencc-by-4.0Dec 2022View details →

Synthetic dataset used for validating MDSPACE method for analyzing continuous conformational variability of biomolecules in cryo-EM single particle images

Synthetic dataset used for validating MDSPACE method for analyzing continuous conformational variability of biomolecules in cryo-EM single particle images. A README file with the contents of the dataset is included.

opencc-by-4.0Apr 2023View details →

Raw Synthetic Particle Image Dataset (RSPID)

Synthetic Particle Image Velocimetry (PIV) data generated by PIV Image Generator Software. Which is a tool that generates synthetic Particle Imaging Velocimetry (PIV) images with the purpose of validating and benchmarking PIV and Optical Flow methods in tracer based imaging for fluid mechanics (Mendes et al., 2020). This data was generated with the following parameters:<ul><li>image width: 665 pixels;</li><li>image height: 630 pixels;</li><li>bit depth: 8 bits;</li><li>particle radius: 1, 2, 3, 4 pixels;</li><li>particle density: 15, 17, 20, 23, 25, 32 particles;</li><li>delta x factor: 0.05, 0.1, 0.15, 0.2, 0.25 %;</li><li>noise level: 1, 5, 10, 15;</li><li>out-of-plane standard deviation: 0.01, 0.025, 0.05;</li><li>flows: rankine uniform, rankine vortex, parabolic, stagnation, shear, decaying vortex.</li></ul>

opencc-by-4.0Mar 2022View details →

deepNIR: Dataset for generating synthetic NIR images and improved fruit detection system using deep learning techniques

In this paper, we present datasets that can be utilised for synthetic near infrared (NIR) image and bounding box level fruit detection system. It is undeniable fact that high-caliber machine learning software frameworks such as Tensorflow or Pytorch and large scale dataset such as ImageNet and COCO, and accelerated GPU hardware support have pushed the limit of machine learning for more than decades. Among these breakthroughs quality dataset is one of important key building blocks that can lead to success in model generalisation and deployment for data-driven deep neural networks. Particularly, synthetic data generation such as generative adversarial networks often requires relatively larger scale data than other supervised approaches. In addition, posing constrains such as geometrical facial constrains in fake face generation or consistent and radiometrically calibrated reflectances from satellite imagery commonly yield better results. We share NIR+RGB dataset that are re-processed from other two public datasets (nirscene and SEN12MS) and our own novel sweetpepper dataset to be able to timely adopt to other following studies. We oversampled from original nirscene dataset at 10, 100, 200, and 400 ratios and total of 127k pair of images. For SEN12MS satellite multispectral dataset, we selected one largest subset; Summer (45k) and All seasons (180k). Our sweetpeppr dataset consists of 1,615 pairs of NIR+RGB images. We demonstrate these NIR+RGB datasets are sufficient to be used for synthetic NIR generation quantitatively and qualitatively. We achieved Frechet Inception Distance (FID) of 11.36, 26.53, and 40.15 for nirscene1, SEN12MS, and sweetpepper dataset respectively. We also release 11 fruits' bounding box annotations that can be exported as various formats using cloud service. 4 newly added fruits [blueberry, cherry, kiwi, and wheat] compounds 11 novel bounding box dastaset together with our previous work in deepFruits project [apple, avocado, capsicum, mango, orange, rockmelon, strawberry]. The total number of bounding box instances is 162k and all bounding box dataset is ready for use from cloud service. For evaluation of these dataset, Yolov5 single stage detector is exploited and reported impressive mean-average-precision, mAP[0.5:0.95] results of [min:0.49, max:0.812]. We hope these dataset is useful and serves as one of baseline for the following up studies.

opencc-by-4.0Jul 2022View details →

A Dataset of Synthetic Images of Outdoor Scenes Taken from Sidewalks, for Temporal Semantic Segmentation Applications

This dataset has been generated using the CARLA simulator (release 0.9.11), an open-source 3D simulator for experiments in autonomous vehicle, based on the Unreal Engine game engine. It comes with pre-made city environment maps. CARLA is distributed with several integrated maps as well as parameters to increase the variety in the dataset. In the release that we have used, there are 13 semantic segmentation classes: None, Building, Fence, Other, Pedestrian, Pole, Lane-marking, Road, Sidewalk, Vegetation, Vehicle, Wall, and Traffic sign. The "None" category corresponds to textures that are not part of an object, such as lawns which are not part of "Vegetation", or sky. In the “Other” category are found objects that are not included in the other classes like plant and flower pots. For smart mobility applications, the “Sidewalks” and “Road” classes are of particular importance to find the way forward, as well as “Buildings” and “Poles” for obstacle avoidance. Sequences are made of 4 images. The dataset is composed of 46436 frames (11609 sequences) partitioned in 41024 frames (10256 sequences) for train, 2696 frames (674 sequences) for validation, and 2716 for test (679 sequences). The size of the images is 800 x 600 (resp. width x height). Additionaly, we have generated another smaller dataset with images taken from 2 different viewpoints: one located on the road and the other located on the sidewalk. The number of frames for train/validation/test is respectively 7288 (1822 sequences) partitioned in 6344 (1687 sequences) for train, 416 frames (104 sequences) for validation, and 424 for test (106 sequences). This smaller dataset is aimed at showing the importance of the viewpoint in the result of semantic segmentation. This can be done by cross-validation: learning on images taken from a viewpoint located on the road and test on images with a viewpoint located on the sidewalk, and vice versa.

opencc-by-4.0Oct 2022View details →

TrueFace: a Dataset for the Detection of Synthetic Face Images from Social Networks

TrueFace is a first dataset of social media processed real and synthetic faces, obtained by the successful StyleGAN generative models, and shared on Facebook, Twitter and Telegram. Images have historically been a universal and cross-cultural communication medium, capable of reaching people of any social background, status or education. Unsurprisingly though, their social impact has often been exploited for malicious purposes, like spreading misinformation and manipulating public opinion. With today's technologies, the possibility to generate highly realistic fakes is within everyone's reach. A major threat derives in particular from the use of synthetically generated faces, which are able to deceive even the most experienced observer. To contrast this fake news phenomenon, researchers have employed artificial intelligence to detect synthetic images by analysing patterns and artifacts introduced by the generative models. However, most online images are subject to repeated sharing operations by social media platforms. Said platforms process uploaded images by applying operations (like compression) that progressively degrade those useful forensic traces, compromising the effectiveness of the developed detectors. To solve the synthetic-vs-real problem "in the wild", more realistic image databases, like TrueFace, are needed to train specialised detectors.

opencc-by-4.0Apr 2023View details →

GAN-based Synthetic VIIRS-like Image Generation over India

Monthly nighttime lights (NTL) can clearly depict an area's prevailing intra-year socio-economic dynamics. The Earth Observation Group at Colorado School of Mines provides monthly NTL products from the Day Night Band (DNB) sensor on board the Visible and Infrared Imaging Suite (VIIRS) satellite (April 2012 onwards) and from Operational Linescan System (OLS) sensor onboard the Defense Meteorological Satellite Program (DMSP) satellites (April 1992 onwards). In the current study, an attempt has been made to generate synthetic monthly VIIRS-like products of 1992-2012, using a deep learning-based image translation network. Initially, the defects of the 216 monthly DMSP images (1992-2013) were corrected to remove geometric errors, background noise, and radiometric errors. Correction on monthly VIIRS imagery to remove background noise and ephemeral lights was done using low and high thresholds. Improved DMSP and corrected VIIRS images from April 2012 - December 2013 are used in a conditional generative adversarial network (cGAN) along with Land Use Land Cover, as auxiliary input, to generate VIIRS-like imagery from 1992-2012. The modelled imagery was aggregated annually and showed an R2 of 0.94 with the results of other annual-scale VIIRS-like imagery products of India, R2 of 0.85 w.r.t GDP and R2 of 0.69 w.r.t population. Regression analysis of the generated VIIRS-like products with the actual VIIRS images for the years 2012 and 2013 over India indicated a good approximation with an R2 of 0.64 and 0.67 respectively, while the spatial density relation depicted an under-estimation of the brightness values by the model at extremely high radiance values with an R2 of 0.56 and 0.53 respectively. Qualitative analysis for also performed on both national and state scales. Visual analysis over 1992-2013 confirms a gradual increase in the brightness of the lights indicating that the cGAN model images closely represent the actual pattern followed by the nighttime lights. Finally, a synthetically generated monthly VIIRS-like product is delivered to the research community which will be useful for studying the changes in socio-economic dynamics over time.

opencc-by-4.0Dec 2023View details →

Rulers2023: An Annotated Dataset of Synthetic and Real Images for Ruler Detection Using Deep Learning

Annotated datasets of synthetic and real ruler images: 1. Synthetic-train 2. Real-train 3. Real-testCitation: Matuzevičius D. Rulers2023: An Annotated Dataset of Synthetic and Real Images for Ruler Detection Using Deep Learning. Electronics. 2023; 12(24):4924. https://doi.org/10.3390/electronics12244924

opencc-by-4.0Mar 2024View details →

Diagnostic and prediction value of synthetic magnetic resonance imaging in acute ischemic stroke patients

This is the supplementary tables for the above paper, mainly including original statistical analysis data.

opencc-by-4.0Apr 2024View details →

PENGWIN Task 2: Pelvic Fragment Segmentation on Synthetic X-ray Images

The <a href="https://pengwin.grand-challenge.org/">PENGWIN segmentation challenge</a> is designed to advance the development of automated pelvic fracture segmentation techniques in both 3D CT scans (Task 1) and 2D X-ray images (Task 2), aiming to enhance their accuarcy and robustness. The full 3D dataset comprises CT scans from 150 patients scheduled for pelvic reduction surgery, collected from multiple institutions using a variety of scanning devices. This dataset represents a diverse range of patient cohorts and fracture types. Ground-truth segmentations for sacrum and hipbone fragments have been semi-automatically annotated and subsequently validated by medical experts, and are available <a href="https://doi.org/10.5281/zenodo.10927452">here</a>. From this 3D data, we have generated high-quality, realistic X-ray images and corresponding 2D labels from the CT data using <a href="https://github.com/arcadelab/deepdrr">DeepDRR</a>, incorporating a range of virtual C-arm camera positions and surgical tools. This dataset contains the training set for fragment segmentation in synthetic X-ray (task 2). The training set is derived from 100 CTs, with 500 images each, for a total of 50,000 training images and segmentations. The C-arm geometry is randomly sampled for each CT within reasonable parameters for a full-size C-arm. The virtual patient is assumed to be in a head-first supine position. Imaging centers are randomly sampled within 50 mm of a fragment, ensuring good visibility. Viewing directions are sampled uniformly on the sphere within 45 degrees of vertical. Half of the images (IDs XXX_0250 - XXX_0500) contain up to 10 simulated K-wires and/or orthopaedic screws oriented randomly in the field of view. The input images are raw intensity images without any windowing or normalization applied. It is standard practice to first apply the negative log transformation and then window each image appropriately for feeding into a model. See the included augmentation pipeline in `pengwin_utils.py` for one approach. For viewing raw images, the <a href="https://imagej.net/software/fiji/">FIJI</a> image viewer is a viable option, but it is recommended to use the included visualization functions in `pengwin_utilities.py` to first apply CLAHE normalization and save to a universally readable PNG (see example usage below). Because X-ray images feature overlapping segmentation maks, the segmentations have been encoded as multi-label uint32 images, where each pixel should be treated as a binary vector with bits 1 - 10 for SA fragments, 11 - 20 for LI, and 21 - 30 for RI. Thus, the raw segmentation files are not viewable with standard image viewing software. `pengwin_utilities.py` includes functions for converting to and from this format and for visualizing masks overlaid onto the original image (see below). To use the utilities, first install dependencies with `pip install -r requirement.txt`. Then, to visualize an image with its segmentation, you can do the following (assuming the training set has been downloaded and unzipped in the same folder): <pre><code>import pengwin_utils from PIL import Image image_path = "train/input/images/x-ray/001_0000.tif" seg_path = "train/output/images/x-ray/001_0000.tif" # load image and masks image = pengwin_utils.load_image(image_path) # raw intensity image masks, category_ids, fragment_ids = pengwin_utils.load_masks(seg_path) # save visualization of image and masks # applies CLAHE normalization to the raw intensity image before overlaying segmentations. vis_image = pengwin_utils.visualize_sample(image, masks, category_ids, fragment_ids) vis_path = "vis_image.png" Image.fromarray(vis_image).save(vis_path) print(f"Wrote visualization to {vis_path}") # Obtain predicted masks, category_ids, and fragment_ids # Category IDs are {"SA": 1, "LI": 2, "RI": 3} # Fragment IDs are the integer labels from label_{category}.nii.gz, with 1 corresponding to the main fragment. pred_masks, pred_category_ids, pred_fragment_ids = masks, category_ids, fragment_ids # replace with your model # save the predicted masks for upload to the challenge # Note: cv2 does not work with uint32 images. It is recommended to use PIL or imageio.v3 pred_seg = pengwin_utils.masks_to_seg(pred_masks, pred_category_ids, pred_fragment_ids) pred_seg_path = "pred/train/output/images/x-ray/001_0000.tif" # ensure dir exists! Image.fromarray(pred_seg).save(pred_seg_path) print(f"Wrote segmentation to {pred_seg_path}")</code></pre> The `pengwin_utils.Dataset` class is provided as an example of a Pytorch dataset, with strong domain randomization included to facilitate sim-to-real performance, but it is recommended to write your own as needed.

opencc-by-4.0Jul 2024View details →

Data pertaining to the published article "Detection of pathological contrast enhancement with synthetic brain imaging from quantitative multiparametric MRI" by Donatelli et al., 2024

Data pertaining to the published article "Detection of pathological contrast enhancement with synthetic brain imaging from quantitative multiparametric MRI" by Donatelli et al., 2024. <a href="https://doi.org/10.1111/jon.13201">https://doi.org/10.1111/jon.13201</a>

opencc-by-4.0Jun 2024View details →

SD4EO: AI-based synthetic satellite Sentinel-2 images of cities and building coverture (RGB+NIR bands)

This dataset has been created as part of the deliverables for ESA’s <a href="https://eo4society.esa.int/projects/sd4eo/">SD4EO project</a>. It consists of synthetic versions of Sentinel-2 images in urban areas. These images were synthetically generated using schematic representations from Open Street Maps as a guide to create AI-based conditioned diffusion model images in the visible and near-infrared spectrum, along with coverage masks for non-residential buildings and the set of residential buildings combined with the former. At least five synthetic variants were generated for each of the eleven cities: <ul> <li>Paris (11 variants)</li> <li>Toulouse (9 variants)</li> <li>Poitiers (8 variants)</li> <li>Bordeaux (6 variants)</li> <li>Limoges (9 variants)</li> <li>Clermont-Ferrand (5 variants)</li> <li>Troyes (6 variants)</li> <li>Le Mans (14 variants)</li> <li>Angers (7 variants)</li> <li>Madrid (15 variants)</li> <li>Niort (6 variants)</li> </ul> The file names within the ZIP archives follow a very simple schema: `assembled_` + city name + usage or band indicator + variant + PNG extension / NC extension Each of the four types of images has a different indicator or band: <ul> <li>`_RGB_` for images encoding visible spectrum signals</li> <li>`_NIR_` for images generated for the near-infrared band</li> <li>`_full_allbuildingmask` for the coverage pixel mask of all building types in floating point</li> <li>`_full_nonresidentialmask` for the coverage pixel mask of non-residential buildings in floating point</li> <li>if we have no indicator, then it is a netCDF file with a labelled xarray that merges RGB+NIR as the original Sentinel-2 spectral bands in full original range</li> </ul> NOTE: This 5th version corrects a minor bug in 3rd version of this dataset. If you want to access to version 4 (with non-already assembled patches), it is also available in the right side control version list. The SD4EO Project is funded by the ESA’s FutureEO programme under contract no. 4000142334/23/I-DT and supervised by ESA Φ-lab.