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.
1,273
datasets available to search
ShareScore release 0.9.0
Dataset results
1,273 results for “Deep Learning”
Deep learning model for characterizing protein-RNA interactions from sequence at single-base resolution
<p> </p> <p><a href="https://zenodo.org/api/records/14021440/draft/files/encode_eclip.h5/content" target="_blank" rel="noopener noreferrer">encode_eclip.h5</a> - This file contains the training, validation, and test data for the Reformer model.</p> <p><a href="https://zenodo.org/api/records/14021440/draft/files/encode_eclip_bc.h5/content" target="_blank" rel="noopener noreferrer">encode_eclip_bc.h5</a> - This file contains the training, validation, and test data for the Reformer-BC model.</p> <p><a href="https://zenodo.org/api/records/14027315/draft/files/Reformer-code.zip/content" target="_blank" rel="noopener">Reformer-code.zip</a> - This file contains the training code of Reformer.</p>
TRANSFORMING BANKING SECURITY: THE ROLE OF DEEP LEARNING IN FRAUD DETECTION SYSTEMS
<p>In the digital banking landscape, the increasing volume of online transactions has heightened the risk of fraudulent activities, necessitating the development of more effective detection systems. This study investigates the efficacy of various machine learning and deep learning algorithms in identifying fraudulent transactions, emphasizing Long Short-Term Memory (LSTM) networks. We implemented and evaluated multiple algorithms, including Logistic Regression, Random Forest, Gradient Boosting Machines (GBM), and XGBoost, on a large-scale credit card transaction dataset. Our results demonstrate that the LSTM model outperforms traditional machine learning algorithms, achieving an accuracy of 98.5%, precision of 87.2%, recall of 85.0%, and an Area Under the Curve (AUC) score of 0.94. These findings highlight the superior capability of LSTM networks to capture complex patterns in sequential transaction data, making them an asset for real-time fraud detection in banking. This research underscores the need for financial institutions to adopt advanced deep learning techniques to enhance their fraud detection systems, thereby minimizing financial losses and improving customer trust.</p>
Data for paper "OC_Finder: A deep learning-based software for osteoclast segmentation, classification, and counting"
<pre>This folder contains the source data for the images we used to train the neural network. Tiff files are microscopic images, and each tiff file has two xls or csv files to indicate the location of osteoclasts or non-osteoclasts in each Tiff file. We performed data collection 3 times, and the data from each data collection is stored separately in 3 folders, "Dataset_1", "Dataset_2", "Dataset_3". The "Dataset_1" folder contains a total of 170 tiff images. They are from the cell culture of wild-type cells stimulated with 50 ng/ml RANKL with/without 100 ng/ml TNF-alpha or 10 ng/ml IL-1beta. The location of cells is recorded in xls files. "IMAGE_NAME osteoclasts.xls" is for the coordination of osteoclasts, and "IMAGE_NAME non-osteoclasts.xls" is for the coordination of non-osteoclasts. The x-coordinate is recorded in the "x" column, and the y-coordinate is in the "y" column, with x=0 and y=0 being the coordinates of the upper left corner of the image. Please ignore the columns other than x and y. The "Dataset_2" folder contains a total of 288 images. They are from the cell culture of wild-type cells or cells with gain-of-function mutation of SH3BP2 (KI) stimulated with 25 or 50 ng/ml of RANKL. The image names are composed of the information about the culture conditions, such as gender of cell sources (female and male), the genotype of the cell source (wt: wild-type, KI: Knock-In mutation in Sh3bp2 resulting in the increased osteoclastogenesis), the concentration of RANKL in the culture media (R25: 25 ng/ml of RANKL, R50: 50 ng/ml of RANKL), and the culture period with RANKL stimulation (day 3: cultured with RANKL for 3 days.), excepting the images with the name starting "Image_". The images with the name Image_NUMBER.tif are from the following culture condition; female KI R25. The location of cells is recorded in xls files. "IMAGE_NAME posi.xls" is for the coordination of osteoclasts, and "IMAGE_NAME nega.xls" is for the coordination of non-osteoclasts. The x-coordinate is recorded in the "x" column, and the y-coordinate is in the "y" column, with x=0 and y=0 being the coordinates of the upper left corner of the image. Please ignore the columns other than x and y. The "additional data" folder contains a total of 288 images, which are identical to the images in the "second batch" folder. Using the same images, we collected additional coordination to increase samples. The location of cells is recorded in csv files. "IMAGE_NAME new posi.xls" is for the coordination of osteoclasts, and "IMAGE_NAME new nega.xls" is for the coordination of non-osteoclasts. The x-coordinate is recorded in the "x" column, and the y-coordinate is in the "y" column, with x=0 and y=0 being the coordinates of the upper left corner of the image. Please ignore the columns other than x and y. Please contact Mizuho Kittaka <mkittaka@iu.edu> for the inquery about the dataset.</pre>
Acoustic video cameras multi-species multi-cameras Training Dataset (TD) for Deep Learning applications
<p>This images dataset, called also TD (Training Dataset), is designed to train/test/validate 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, formatted in the yolo-darknet format as explained in <a href="https://github.com/AlexeyAB/darknet">https://github.com/AlexeyAB/darknet</a>, includes also a large number of images without any fish passage, in order to test the effect of negative data on the trainings.</p>
Complete results of Procedure section in manuscript "PathML: A unified framework for whole-slide image analysis with deep learning"
<p>The results from a complete run of the Procedure section of the paper "PathML: A unified framework for whole-slide image analysis with deep learning".</p>
Data for "Deep learning-based model for diagnosing Alzheimer's disease and tauopathies"
<p>Image datasets and tuned models used in the paper (Koga et al., 2021). Data.zip contains image and text files for training models. Test.zip contains 12 images from 4 patients, which are a part of the hold-out dataset images used in the paper. There are 9 CSV files, which contain the results of tau burden quantification. Python code is available at GitHub (<a href="https://github.com/Koga-MD/DL-Tauopathies">https://github.com/Koga-MD/DL-Tauopathies</a>). </p>
Experimental data for "DeepMetis: Augmenting a Deep Learning Test Set to Increase its Mutation Score" paper
<p>Experimental data for "DeepMetis: Augmenting a Deep Learning Test Set to Increase its Mutation Score" paper</p>
Dataset for deep-learning in-situ classification of HIV-1 virion morphology
<p>This dataset contains TEM micrographs for HIV-1 virion samples intended for classification and detection as follows:</p> <ol> <li> <p>HIV-1_virion_classification_backbone_dataset.zip : Contains 1806 .tif images of isolated HIV-1 virions extracted and augmented from TEM micrographs. The images are divided into training (1443 images) and validation (363 images) sets and each of these is divided into eccentric, mature, immature labeled folders:</p> <ul> <li> <p>HIV-1_virion_classification_backbone_dataset/</p> <ul> <li> <p>train/</p> <ul> <li> <p>eccentric/</p> </li> <li> <p>immature/</p> </li> <li> <p>mature/</p> </li> </ul> </li> <li> <p>val/</p> <ul> <li> <p>eccentric/</p> </li> <li> <p>immature/</p> </li> <li> <p>mature/</p> </li> </ul> </li> </ul> </li> </ul> </li> <li> <p>HIV-1_rcnn_dataset_full.zip : Contains 59 .tif TEM micrographs of HIV-1 samples as well as a matching .csv file recording the attributes of each viral instance and coordinates of the rectangular region that contains it:</p> </li> </ol> <ul> <li> <p>region_data_<image_id>.csv:</p> <ul> <li> <p>filename: Name of the image this csv refers to. (Ex: 0131001.png)</p> </li> <li> <p>file_size: Size (bytes) of the image this csv refers to. (Ex: 11755590)</p> </li> <li> <p>file_attributes: Specific attributes of the micrograph. (Ex: None)</p> </li> <li> <p>region_count: Number of viral instances detected in the micrograph. (Ex: 39)</p> </li> <li> <p>region_id: ID of a specific viral region. (Ex: 1)</p> </li> <li> <p>region_shape_attributes: Coordinates of the bounding box of <region_id> that contains a virion. (Ex: {"name":"rect","x":1022,"y":357,"width":225,"height":228})</p> </li> <li> <p>region_attributes: Classification of the virion enclosed in this region (eccentric/mature/immature). (Ex: {"particle_class":"mature"})</p> </li> </ul> </li> </ul> <p>The images are divided into training (46 images) and validation (13 images) sets and each of these contains folder for each micrograph:</p> <ul> <li> <p>HIV-1_rcnn_dataset_full/</p> <ul> <li> <p>train/</p> <ul> <li> <p>0131001/</p> <ul> <li> <p>0131001.png</p> </li> <li> <p>region_data_0131001.csv</p> </li> </ul> </li> <li> <p>0131004/</p> <ul> <li> <p>0131004.png</p> </li> <li> <p>region_data_0131004.csv</p> </li> </ul> </li> <li> <p>…</p> </li> </ul> </li> <li> <p>val/</p> <ul> <li> <p>0131002/</p> <ul> <li> <p>0131002.png</p> </li> <li> <p>region_data_0131002.csv</p> </li> </ul> </li> <li> <p>0131003/</p> <ul> <li> <p>0131003.png</p> </li> <li> <p>region_data_0131003.csv</p> </li> </ul> </li> <li> <p>…</p> </li> </ul> </li> </ul> </li> </ul> <p>The first dataset (HIV-1_virion_classification_backbone_dataset.zip) is intended for viral classification algorithms while the second dataset (HIV-1_rcnn_dataset_full) is intended for detection and segmentation algorithms (for example RCNN).</p> <p>Applications of this dataset as well as source code can be found at <a href="https://github.com/Perilla-lab/TEMNet">https://github.com/Perilla-lab/TEMNet</a> .</p>
Supplement: Multicenter validated detection of focal cortical dysplasia using deep learning
<p><span><b><span>Objective</span></b><span>. </span><span><span>To test the hypothesis that a multicenter-validated computer deep learning algorithm detects MRI-negative focal cortical dysplasia (FCD).</span></span></span></p> <p><span><b><span>Methods</span></b><span>. We used clinically acquired 3D T1-weighted and 3D FLAIR MRI of 148 patients (median age, 23 years [range, 2-55]; 47% female) with histologically verified FCD at nine centers to train a deep convolutional neural network (CNN) classifier. Images were initially deemed as MRI-negative in 51% of cases, in whom intracranial EEG determined the focus. For risk stratification, the CNN incorporated Bayesian uncertainty estimation as a measure of confidence. To evaluate performance, detection maps were compared to expert FCD manual labels. </span><span><span>We also tested sensitivity in an independent cohort of </span></span><span><span>23 FCD cases (13±10 years</span></span><span><span>).</span></span><span> Applying the algorithm to 38 healthy and 63 temporal lobe epilepsy disease controls tested specificity. </span></span></p> <p><span><b><span>Results.</span></b><span> Overall sensitivity was 93% (137/148 FCD detected) using a leave-one-site-out cross-validation, with an average of six false positives per patient. Sensitivity in MRI-negative FCD was 85%. In 73% of patients, the FCD was among the clusters with the highest confidence; in half it ranked the highest. </span><span><span>Sensitivity in the </span></span><span><span>independent cohort was </span></span><span><span>83% (19/23; average of five false positives per patient).</span></span><span> Specificity was 89% in healthy and disease controls.</span></span></p> <p><span><b><span>Conclusions</span></b><span>. This first multicenter-validated deep learning detection algorithm yields the highest sensitivity to date in MRI-negative FCD. By pairing predictions with risk stratification</span><span><span> this classifier may assist clinicians to adjust hypotheses relative to other tests, increasing </span></span><span>diagnostic confidence</span><span><span>.</span></span><span> Moreover, generalizability across age and MRI hardware </span><span><span>makes this approach ideal for pre-surgical evaluation of MRI-negative epilepsy. </span></span></span></p> <p><b><span>Classification of evidence</span></b><span>. This study provides Class III evidence that deep learning on multimodal MRI accurately identifies FCD in epilepsy patients initially diagnosed as MRI-negative.</span></p>
Supplementary data for "Deep learning for industrial processes: Forecasting amine emissions from a carbon capture plant"
<p>A preliminary analysis of the data already has been discussed in <a href="https://dx.doi.org/10.2139/ssrn.3812299">10.2139/ssrn.3812299</a>.</p> <p><strong>Raw data</strong></p> <p>Raw measurement data is in the Excel files `day*_raw.xlsx`.</p> <p><strong>Model</strong></p> <p>Covariate and label scaler objects are serialized in joblib format in the following files:</p> <ul> <li>20210812_y_transformer_co2_ammonia_reduced_feature_set</li> <li>20210812_y_transformer__reduced_feature_set</li> <li>20210812_x_scaler_reduced_feature_set</li> </ul> <p>Checkpoints of the models are in the `*.pth.tar` files. An example for loading the models is:</p> <pre><code class="language-python">from pyprocessta.model.tcn import TCNModelDropout model_cov = TCNModelDropout( input_chunk_length=8, output_chunk_length=1, num_layers=5, num_filters=16, kernel_size=6, dropout=0.3, weight_norm=True, batch_size=32, n_epochs=100, log_tensorboard=True, optimizer_kwargs={"lr": 2e-4}, ) model_cov.load_from_checkpoint('20210814_2amp_pip_model_reduced_feature_set_darts')</code></pre> <p>which assumes that the checkpoints are placed as `model_best.pth.tar` in a folder called `20210812_2amp_pip_model_reduced_feature_set_darts`.</p> <p> </p>
CNN models and training, validation and test datasets for "PlotMI: interpretation of pairwise interactions and positional preferences learned by a deep learning model from sequence data"
<p>Convolutional neural network (CNN) models and their respective training, validation and test datasets used in manuscript:</p> <p>Tuomo Hartonen, Teemu Kivioja and Jussi Taipale, "PlotMI: interpretation of pairwise interactions and positional preferences learned by a deep learning model from sequence data"</p>
NGS Data Accompanying "Deep Learning Enables Design of Multifunctional Synthetic Human Gut Microbiome Dynamics"
<p>NGS Data Accompanying "Deep Learning Enables Design of Multifunctional Synthetic Human Gut Microbiome Dynamics", currently in review.</p>
Assessment of Non-Invasive Blood Pressure Prediction from PPG and rPPG Signals Using Deep Learning
<p>This dataset is a subset of the MIMIC-III dataset used for non-invasive blood pressure prediction. PPG and ABP data were divided into windows of 7s length (875 data points). Systolic and diastolic blood pressure values were derived from the ABP windows. Each sample of the dataset consists of a PPG signal and blood pressure values as well as a unique subject identifier. The file consists of three datasets:</p> <ul> <li>PPG: PPG data of size 905,400 x 875</li> <li>label: BP data of size 905,400 x 2</li> <li>subject_idx: subject affiliation of each sample (size 905,400 x 1)</li> </ul> <p>Furthermore, this submission contains the following models:</p> <ul> <li>AlexNet</li> <li>ResNet50</li> <li>LSTM</li> <li>Architecture published by Slapnicar et al. 2019</li> </ul> <p>The architectures were trained using a non-mixed dataset derived from the MIMIC-III waveform database. Samples were divided between training, validation and test set based on their subject affiliation preventing contamination of validation and test sets with samples from subjects used for training.</p>
A deep learning dataset for underwater object detection of tropical freshwater fish species in northern Australia
<p>This dataset includes 44,112 images with 82,904 bounding box annotations for 23 tropical freshwater fish taxa from northern Australia. </p> <p>Images were derived from Remote Underwater Video (RUV) deployments in deep channel and shallow lowland billabongs, Kakadu National Park, Northern Territory Australia. RUV deployments were conducted during the <a href="https://www.dcceew.gov.au/science-research/supervising-scientist">Supervising Scientists</a> annual fish monitoring program in the 2016, 2017 and 2018 recessional flow period (dry season). More information can be found <a href="https://www.dcceew.gov.au/sites/default/files/documents/ss-atr-2020-21.pdf">here</a>.</p> <ul> <li>All images are in .jpg format and are 1920x1080 in dimension.</li> <li>Bounding box annotations are in COCO format. </li> </ul> <p>Two .zip files are included:</p> <ul> <li><a href="https://zenodo.org/api/files/990412db-e633-4f82-9b32-6990ef439ccd/202210-KakaduFishAI-CompactModel.zip">202210-KakaduFishAI-CompactModel.zip</a>: includes compact model weights in tensorflow format (.pb) trained using Azure's Custom Vision platform. This model is suitable for edge devices due to its reduced size. Code is provided to use the compact model for inferencing. </li> <li> <a href="https://zenodo.org/api/files/990412db-e633-4f82-9b32-6990ef439ccd/202210-KakaduFishAI-TrainingData.zip">202210-KakaduFishAI-TrainingData.zip</a>: includes all images and one COCO (.json) file with annotations. </li> </ul> <p>Fish taxa include: </p> <ol> <li><em>Ambassis agrammus</em></li> <li><em>Ambassis macleayi</em></li> <li><em>Amniataba percoides</em></li> <li><em>Craterocephalus stercusmuscarum</em></li> <li><em>Denariusa bandata</em></li> <li><em>Glossamia aprion</em></li> <li><em>Glossogobius</em> spp.</li> <li><em>Hephaestus fuliginosus</em></li> <li><em>Lates calcarifer</em></li> <li><em>Leiopotherapon unicolor</em></li> <li><em>Liza ordensis</em></li> <li><em>Megalops cyprinoides</em></li> <li><em>Melanotaenia nigrans</em></li> <li><em>Melanotaenia splendida inornata</em></li> <li><em>Mogurnda mogurnda</em></li> <li><em>Nemetalosa erebi</em></li> <li><em>Neoarius</em> spp.</li> <li><em>Neosilurus</em> spp.</li> <li><em>Oxyeleotris</em> spp.</li> <li><em>Scleropages jardinii</em></li> <li><em>Strongylura kreffti</em></li> <li><em>Syncomistes butleri</em></li> <li><em>Toxotes chatareus</em></li> </ol> <p>If you use this data for your own deep learning project we'd love to hear about how you used this dataset: andrew.jansen@environment.gov.au.</p>
Training Deep Learning Models to Estimate SWAT Parameters using Streamflow Observations
<p>This folder provides the simulation and observational data</p> <p>Simulation data using SWAT (1000 realz)<br> Train, Val, and Test splits (80/10/10)<br> Observational data for ARW (WY2000-2016)</p>
Identification of free-ranging mugger crocodiles by applying deep learning methods on UAV imagery
<p>Individual identification contributes significantly towards investigating behavioral mechanisms of animals and understanding underlying ecological principles. Most studies employ invasive procedures for individually identifying organisms. In recent times, computer-vision techniques have served as an alternative to invasive methods. However, these studies primarily rely on user input data collected from captivity or from individuals under partially restrained conditions. Challenges in collecting data from free-ranging individuals are higher when compared to captive populations. However, the former is a far more important priority for real-world applications. In this paper, we used UAV to collect data from free-ranging mugger crocodiles <em>Crocodylus</em> <em>palustris</em>. We applied convolutional neural networks (CNNs) to individually identify muggers based on their dorsal scute patterns. The CNN model was trained on a data set of 88,000 images focusing on the mugger's dorsal body. The data was collected from 143 individuals across 19 different locations along the western part of India. We trained two CNN models, one with an annotated bounding box approach, the YOLO-v5l, and another without annotations, the Inception-v3. We used two parameters, True Positive Rate (TPR) and True Negative Rate (TNR), to validate the efficiency of the trained models. Using YOLO-v5l, TPR (re-identification of trained muggers) and TNR (differentiating untrained muggers as 'unknown') values at 0.84 threshold were 88.8% and 89.6%, respectively. The trained model showed 100% TNR for the non-mugger species, the Gharial <em>Gavialis</em> <em>gangeticus</em>, and the Saltwater crocodile <em>Crocodylus</em> <em>porosus</em>. The performance of the CNN model was reliable and accurate while using only 125 images per individual for training purposes. Inception-v3 underperformed for both the parameters, thus, showing that a bounding box approach (YOLO-v5l model) with background elimination is a promising method to individually identify free-ranging mugger crocodiles. Our manuscript demonstrates that UAV imagery appears to be a promising tool for non-invasive collection of data from free-ranging populations. It can be used to train open-source algorithms for individual identification. Further, the identification method is entirely based upon dorsal scute patterns, which can be applied to different crocodilian species, as well.</p>
Simulated industrial CT dataset for deep learning with dual-energy tomograms and ground truth material maps for copper and iron
<p>We use this dataset for training and evaluation of a deep learning model to discriminate multi-material systems with X-ray CT.</p> <p>The dataset consists of:</p> <ul> <li>inputs: dual-energy tomograms as binary files without a header (tensor <strong>shape for numpy: 2x128x128 @float32</strong>) <ul> <li>simulated spectra are 250kVp and 450kVp both prefiltered using 2mmCuSn</li> </ul> </li> <li>outputs: the material maps a.k.a. ground truths for the training (same shape as inputs) <ul> <li>sampled with a delaunay algorithm and randomly filled with iron and copper fractions</li> </ul> </li> </ul> <p>The <strong>dataset is normalized to [0, 1]</strong>, so you have to multiply by the mass densities of copper and iron to obtain effective fractions in g/cm^3.</p>
Benchmark datasets for detection and identification of insects from camera trap images with deep learning
<p><strong>Insect benchmark datasets for training, validation and test (train1201.zip, val1201.zip and test1201.zip) with time-lapse images as described in paper:</strong></p> <p><a href="https://www.biorxiv.org/content/10.1101/2022.10.25.513484v1">Bjerge K, Alison J, Dyrmann M, Frigaard C.E., Mann H. M. R., Høye T.T., Accurate detection and identification of insects from camera trap images with deep learning, bioRxiv:10.1101/2022.10.25.513484v1</a></p> <p>Labels in <strong>YOLO format: <a href="https://github.com/ultralytics/yolov5/issues/2293">ultralytics/yolov5: label format</a></strong></p> <p>The annotated training and validation datasets contains insects of nine different species as listed below:</p> <table> <tbody> <tr> <td>0 <em>Coccinellidae septempunctata</em></td> </tr> <tr> <td>1 <em>Apis mellifera</em></td> </tr> <tr> <td>2 <em>Bombus lapidarius</em></td> </tr> <tr> <td>3 <em>Bombus terrestris</em></td> </tr> <tr> <td>4 <em>Eupeodes corolla</em></td> </tr> <tr> <td>5 <em>Episyrphus balteatus</em></td> </tr> <tr> <td>6 <em>Aglais urticae</em></td> </tr> <tr> <td>7 <em>Vespula vulgaris</em></td> </tr> <tr> <td>8 <em>Eristalis tenax</em></td> </tr> </tbody> </table> <p>The test dataset contains additional classes of insects.</p> <table> <tbody> <tr> <td>9 Non-Bombus Anthophila</td> </tr> <tr> <td>10 Bombus spp.</td> </tr> <tr> <td>11 Syrphidae</td> </tr> <tr> <td>12 Fly spp.</td> </tr> <tr> <td>13 Unclear insect</td> </tr> <tr> <td>14 Mixed animals:<br> ——————————<br> Rhopalocera<br> Non-Anthophila Hymenoptera<br> Non-Syrphidae Diptera<br> Non-Conccinalidae Coleoptera<br> Concinellidae<br> Other animals</td> </tr> </tbody> </table> <p><strong>There are two naming conventions for image (.jpg) and label (.txt) files.</strong></p> <p><em>Background images without insects are named</em>:<br> “<strong>X_Seq-YYYYMMDDHHMMSS</strong>-snapshot”.<br> E.g.:<br> Background image: 12_13-20190704172200-snapshot.jpg<br> Empty label file: 12_13-20190704172200-snapshot.txt</p> <p><em>Images annotated with insects are named:</em><br> “<strong>SZ_IP-MonthDate_C_Seq-YYYYMMDDHHMMSS</strong>”.<br> E.g.:<br> Image file: S1_146-Aug23_1_156-20190822133230.jpg<br> Label file: S1_146-Aug23_1_156-20190822133230.txt</p> <p><strong>Abbreviations</strong>:</p> <p><strong>YYYYMMDDHHMMSS </strong>– Capture timestamp with year, month, date, hour, minutes, and second<br> <strong>Seq</strong> – Sequence number created by the motion program to separate images<br> <strong>C</strong> – Identification of two cameras with Id=0 or Id=1 in system identified by <strong>SZ_IP</strong><br> <strong>MonthDate </strong>– Folder name for where the original image were stored in the system<br> <strong>SZ_IP</strong> – Identification of five camera systems: S1_123, S2_146, S3_194, S4_199, S5_187 (Two cameras in each system)<br> <strong>X</strong> – An index number related to a specific camera and folder ensuring unique file names of background images from different camera systems.<br> <br> The important information in a filename is system (<strong>SZ_IP</strong>), camera Id (<strong>C</strong>) and timestamp (<strong>YYYYMMDDHHMMSS</strong>).</p> <p><strong>The three best YOLOv5 models (YOLOv5models.zip) from the paper are available in pytorch format.</strong></p> <p>All models are tested with YOLOv5 release v7.0 (22-11-2022): <a href="https://github.com/ultralytics/yolov5">ultralytics/yolov5: YOLOv5 in PyTorch</a></p> <p><strong>insect1201-bestF1-640v5m.pt</strong>: Model no. 6 in Table 2 (F1=0.912)<br> <strong>insect1201-bestF1-1280v5m6.pt</strong>: Model no. 8 in Table 2 (F1=0.925)<br> <strong>insect1201-bestF1-1280v5m6.pt</strong>: Model no. 10 in Table 2 (F1=0.932)</p> <p><strong>insects-1201val.yaml</strong>: YAML file with label names to train YOLOv5</p> <p><strong>trainInsects-1201m.sh</strong>: Linux bash shell script with parameters to train YOLOv5m6<br> <strong>valInsectsF1-1201.sh</strong>: Linux bash shell script with parameters to validated models</p> <p> </p>
Yield Prediction Through Integration of Genetic, Environment, and Management Data Through Deep Learning: Cleaned Data
<p>The included files and script are to allow for reconstruction of the data directory and cleaned data used in "Yield Prediction Through Integration of Genetic, Environment, and Management Data Through Deep Learning" ( https://doi.org/10.1101/2022.07.29.502051 ). Code used is available at 10.5281/zenodo.7401113 .</p> <table> <tbody> <tr> <th>Filename</th> <th>Description</th> </tr> <tr> <td>interim.tar.gz</td> <td>Contains site grouping dictonary</td> </tr> <tr> <td>processed.tar.gz</td> <td>Processed data</td> </tr> <tr> <td>raw.tar.gz</td> <td>Input data</td> </tr> <tr> <td>SetupInstructions.sh</td> <td>Bash script to prepare folders and unzipped data expected by code in 10.5281/zenodo.7401113</td> </tr> <tr> <td>SetupInstructions.txt</td> <td>Instructions for unzipping the data</td> </tr> <tr> <td>Train_Test_Split_Reference_Phenotypes.csv</td> <td>Reference spreadsheet to allow for easily exploring training and test set groupings</td> </tr> </tbody> </table> <ul> </ul> <p>This work was supported through funding from the USDA Agricultural Research Service, ARS project number 5070-21000-041-000-D. Raw data provided by the [Genomes to Field Initiative](https://www.genomes2fields.org/) and the [Daymet database](https://daymet.ornl.gov/).</p>
DeepARG: a deep learning approach for predicting antibiotic resistance genes from metagenomic data
<p>Growing concerns about increasing rates of antibiotic resistance call for expanded and comprehensive global monitoring. Advancing methods for monitoring of environmental media (e.g., wastewater, agricultural waste, food, and water) is especially needed for identifying potential resources of novel antibiotic resistance genes (ARGs), hot spots for gene exchange, and as pathways for the spread of ARGs and human exposure. Next-generation sequencing now enables direct access and profiling of the total metagenomic DNA pool, where ARGs are typically identified or predicted based on the “best hits” of sequence searches against existing databases. Unfortunately, this approach produces a high rate of false negatives. To address such limitations, we propose here a deep learning approach, taking into account a dissimilarity matrix created using all known categories of ARGs. Two deep learning models, DeepARG-SS and DeepARG-LS, were constructed for short read sequences and full gene length sequences, respectively. Evaluation of the deep learning models over 30 antibiotic resistance categories demonstrates that the DeepARG models can predict ARGs with both high precision (> 0.97) and recall (> 0.90). The models displayed an advantage over the typical best hit approach, yielding consistently lower false negative rates and thus higher overall recall (> 0.9). As more data become available for under-represented ARG categories, the DeepARG models’ performance can be expected to be further enhanced due to the nature of the underlying neural networks. Our newly developed ARG database, DeepARG-DB, encompasses ARGs predicted with a high degree of confidence and extensive manual inspection, greatly expanding current ARG repositories. The deep learning models developed here offer more accurate antimicrobial resistance annotation relative to current bioinformatics practice. DeepARG does not require strict cutoffs, which enables identification of a much broader diversity of ARGs. The DeepARG models and database are available as a command line version and as a Web service at <a href="http://bench.cs.vt.edu/deeparg">http://bench.cs.vt.edu/deeparg</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.