Find research datasets worth reusing

opencc-by-4.0Apr 2023View details →

Uncovering local aggregated air quality index with smartphone captured images leveraging efficient deep convolutional neural network

Short Description: In this research, we vigorously analyze the difficulties of predicting location-specific PM2.5 concentration from photos captured by smartphone cameras. Here, we particularly focus on Dhaka, the capital of Bangladesh, considering its very high level of air pollution exposure to a huge number of its dwellers. In our research, we develop a Deep Convolutional Neural Network (DCNN) and train it using more than a thousand outdoor photos captured and labeled by us. We capture the photos at various locations in Dhaka, Bangladesh, and label them based on PM2.5 concentration data extracted from the local US consulate as computed by the NowCast algorithm. During training with the dataset, our model learns a correlation index through supervised learning, which improves the model's ability to act as a Picture-based Predictor of PM2.5 Concentration (PPPC) making it capable of detecting comparable daily aggregated AQI index from a photo captured by a smartphone. Code and More Details: https://github.com/lepotatoguy/aqi

opencc-by-4.0May 2023View details →

Interpretation of EKG with Image Recognition and Convolutional Neural Networks

Dataset used in the training and evaluation of "Interpretation of EKG with Image Recognition and Convolutional Neural Networks" Abstract: Electrocardiograms (EKG) form the backbone of all cardiovascular diagnosis, treatment and follow up. Given the pivotal role it plays in modern medicine, there have been multiple efforts to computerize the EKG interpretation with algorithms to improve efficiency and accuracy. Unfortunately, many of these algorithms are machine specific and run-on proprietary signals generated by that machine, hence not generalizable. We propose the development of an image recognition model which can be used to read standard EKG strips. A convolutional neural network (CNN) was trained to classify 12-lead EKGs between 7 clinically important diagnostic classes. An austere variation of the MobileNetV3 model was trained from the ground up on publicly available labeled training set. The precision per class varies from 52% to 91%. This is a novel approach to EKG interpretation as an image recognition problem.

opencc-by-4.0Aug 2023View details →

Prediction of Spheroid Cell Death using Fluorescence Staining and Convolutional Neural Networks

This repository contains training, validation, and testing of fluorescence image data sets with their label for spheroid cell death classification.  These data are intended to be used in the paper "Prediction of Spheroid Cell Death using Fluorescence Staining and Convolutional Neural Networks" currently submitted

opencc-by-4.0Oct 2023View details →

Data --- "Optimization of Convolutional Neural Network models for spatially coherent multi-site fire danger predictions"

Data to reproduce the results of the manuscript entitled "Optimization of Convolutional Neural Network models for spatially coherent multi-site fire danger predictions" submitted to Geophysical Research Letters. The companion jupyter notebook can be found in DOI: <a href="https://doi.org/10.5281/zenodo.8387558">10.5281/zenodo.8387558</a>

publicDec 2024View details →

Data from: Dorsoventral comparison of intraspecific variations in the butterfly wing pattern using a convolutional neural network

Open the record for dataset details and reuse information.

publicJan 2022View details →

Using convolutional neural networks to efficiently extract immense phenological data from community science images

Open the record for dataset details and reuse information.

publicFeb 2025View details →

Data from: Convolutional neural networks trained on internal variability predict forced response of TOA radiation by learning the pattern effect

Open the record for dataset details and reuse information.

publicJan 2021View details →

Data from: A reusable pipeline for large-scale fiber segmentation on unidirectional fiber beds using fully convolutional neural networks

Open the record for dataset details and reuse information.

dryad36/100

MVCNN++: CAD model shape classification and retrieval using multi-view convolutional neural networks

Deep neural networks have shown promising success towards the classification and retrieval tasks for images and text data. While there have been several implementations of deep networks in the area of computer graphics, these algorithms do not translate easily across different datasets, especially for shapes used in product design and manufacturing domain. Unlike datasets used in the 3D shape classification and retrieval in the computer graphics domain, engineering level description of 3D models do not yield themselves to neat distinct classes. The current study looks at an improved form of the 3D shape deep learning algorithm for classification and retrieval through the use of techniques such as relaxed classification, use of prime angled camera angles for capturing feature detail and transfer learning for reducing the amount of data and processing time needed to train shape recognition algorithms. The proposed algorithm (MVCNN++) builds on top of multi-view convolutional neural network (MVCNN) algorithm, improving its efficacy for manufacturing part classification by enabling use of part metadata, yielding an improvement of almost 6% over the original version. With the explosive growth of 3D product models available in publicly available repositories, search and discovery of relevant models is critical to democratizing access to design models.

opencc-zeroAug 2020View details →

opencc-by-4.0Sep 2020View details →

Meaning maps and saliency models based on deep convolutional neural networks are insensitive to image meaning when predicting human fixations - data

Data from the paper: Meaning maps and saliency models based on deep convolutional neural networks are insensitive to image meaning when predicting human fixations. Preprint: https://www.biorxiv.org/content/10.1101/840256v1 Marek A. Pedziwiatr marek.pedziwi@gmail.com September 2020

opencc-by-4.0Nov 2020View details →

Recognized trophoblast-like cells conversion from human embryonic stem cells by BMP4 based on convolutional neural network

The use of models of stem cell differentiation to trophoblastic cells provides an effective perspective for understanding the early molecular events in the establishment and maintenance of human pregnancy. In combination with the newly developed deep learning technology, the automated identification of this process can greatly accelerate the contribution to relevant knowledge. Based on the transfer learning technique, we used a convolutional neural network to distinguish the microscopic images of Embryonic stem cells (ESCs) from differentiated trophoblasts -like cells (TBL). To tackle the problem of insufficient training data, the strategies of data augmentation were used. The results showed that the convolutional neural network could successfully recognize trophoblast cells and stem cells automatically, but could not distinguish TBL from the immortalized trophoblast cell lines in vitro (JEG-3 and HTR8-SVneo). We compare the recognition effect of the commonly used convolutional neural network, including DenseNet, VGG16, VGG19, InceptionV3, and Xception. This study extends the deep learning technique to trophoblast cell phenotype classification and paves the way for automatic bright-field microscopic image analysis of trophoblast cells in the future.

opencc-by-4.0Jan 2021View details →

Convolutional Neural Network Formulation to Compare 4D Seismic and Reservoir Simulation Models

This dataset contains the .npy (numpy) files of the simulation models and reference discussed in the paper "Convolutional Neural Network Formulation to Compare 4D Seismic and Reservoir Simulation Models". The folders contain all simulation models and reference maps already divided in subregions. Each .npy file is a numpy 2D array with delta IP or delta Sw values. The csv files contain the 3-tuples and the selected model in each. There are two csv files: the first is the dataset used for training the CNN, with 1280 labeled tuples evaluated by a single specialist. The second is the ground-truth, with 164 tuples evaluated by three specialists (in which 2 or more agreed on the selected model), used for validating the models and comparing different approaches. We also provide a Python code to read and visualize the .npy files.

dryad36/100

Data from: A convolutional neural network for detecting sea turtles in drone imagery

1. Marine megafauna are difficult to observe and count because many species travel widely and spend large amounts of time submerged. As such, management programs seeking to conserve these species are often hampered by limited information about population levels. 2. Unoccupied aircraft systems (UAS, aka drones) provide a potentially useful technique for assessing marine animal populations, but a central challenge lies in analyzing the vast amounts of data generated in the images or video acquired during each flight. Neural networks are emerging as a powerful tool for automating object detection across data domains and can be applied to UAS imagery to generate new population-level insights. To explore the utility of these emerging technologies in a challenging field setting, we used neural networks to enumerate olive ridley turtles (Lepidochelys olivacea) in drone images acquired during a mass-nesting event on the coast of Ostional, Costa Rica. 3. Results revealed substantial promise for this approach; specifically, our model detected 8% more turtles than manual counts while effectively reducing the manual validation burden from 2,971,554 to 44,822 image windows. Our detection pipeline was trained on a relatively small set of turtle examples (N=944), implying that this method can be easily bootstrapped for other applications, and is practical with real-world UAS datasets. 4. Our findings highlight the feasibility of combining UAS and neural networks to estimate population levels of diverse marine animals and suggest that the automation inherent in these techniques will soon permit monitoring over spatial and temporal scales that would previously have been impractical.

opencc-zeroDec 2018View details →

opencc-by-4.0Nov 2023View details →

Robustness assessment of a C++ implementation of the LeNet-5 convolutional neural network.

The architecture of the LeNet-5 convolutional neural network (CNN) was defined by LeCun in its paper "Gradient-based learning applied to document recognition" (https://ieeexplore.ieee.org/document/726791) to classify images of hand written digits (MNIST dataset).This architecture has been customized to use Rectified Linear Unit (ReLU) as activation functions instead of Sigmoid.It consists of the following layers:<ul><li>conv1: Convolution 2D, 1 input channel (28x28), 3 output channels (28x28), kernel size 5, stride 1, padding 2.</li><li>relu1: Rectified Linear Unit (3@28x28).</li><li>max1: Subsampling buy max pooling (3@14x14).</li><li>conv2: Convolution 2D, 3 input channels (14x14), 6 output channels (14x14), kernel size 5, stride 1, padding 2.</li><li>relu2: Rectified Linear Unit (6@14x14).</li><li>max2: Subsampling buy max pooling (6@7x7).</li><li>fc1: Fully connected (294, 147)</li><li>fc2: Fully connected (147, 10)</li></ul>The fault hypotheses for this work include the occurrence of:<ul><li>BF: single bit-flip faults</li><li>S0: single, double-adjacent and triple-adjacent stuck-at-0 faults</li><li>S1: single, double-adjacent and triple-adjacent stuck-at-1 faults</li></ul>In the memory cells containing all the parameters of the CNN:  <ul><li>w: weights (float32)</li><li>b: biases (float32)</li></ul>Images 200 to 249 from the MNIST dataset have been used as workload.This dataset contains the raw data obtained from running exhaustive fault injection campaigns for all considered fault models, targeting all considered locations and for all the images in the workload.<h3>Files information</h3><ul><li>golden_run.csv: Prediction obtained for all the images considered in the workload in the absence of faults (Golden Run). This is intended to act as oracle to determine the impact of injected faults.</li><li>single_faults/bit_flip folder: Prediction obtained for all the images considered in the workload in presence of single bit-flip faults. There is one file for each parameter of each layer.</li><li>single_faults/stuck_at_0 folder: Prediction obtained for all the images considered in the workload in presence of single stuck-at-0 faults. There is one file for each parameter of each layer.</li><li>single_faults/stuck_at_1 folder: Prediction obtained for all the images considered in the workload in presence of single stuck-at-1 faults. There is one file for each parameter of each layer.</li><li>double_adjacent_faults/stuck_at_0 folder: Prediction obtained for all the images considered in the workload in presence of double adjacent stuck-at-0 faults. There is one file for each parameter of each layer.</li><li>double_adjacent_faults/stuck_at_1 folder: Prediction obtained for all the images considered in the workload in presence of double adjacent stuck-at-1 faults. There is one file for each parameter of each layer.</li><li>triple_adjacent_faults/stuck_at_0 folder: Prediction obtained for all the images considered in the workload in presence of triple adjacent stuck-at-0 faults. There is one file for each parameter of each layer.</li><li>triple_adjacent_faults/stuck_at_1 folder: Prediction obtained for all the images considered in the workload in presence of triple adjacent stuck-at-1 faults. There is one file for each parameter of each layer.</li></ul><h3>Methodology information</h3>First, the CNN was used to classify all the images of the workload in the absence of faults to get a reference to determine the impact of faults. This is golden_run.csv file.After that, one fault injection experiment was executed for each of the 16 most significant bits bit of each element of each parameter of the CNN, as previous fault injection experiments showed that the occurrence of the considered faults in the 16 least significant bits does not impact the behaviour of the network.Each experiment consisted in:<ul><li>Affecting the bits (inverting it in case of bit-flip faults, setting it to 0 or 1 in case of stuck-at-0 or atuck-at-1 faults) identified by the mask.</li><li>Classifying all the images of the workload in the presence of this fault. The obtained output was stored in a given .csv file.</li><li>Removing the fault from the CNN by restoring the affected bits to its previous value.</li></ul><h3>List of variables (Name : Description (Possible values))</h3><ul><li>IMGID: Integer number identifying the considered image (200-249).</li><li>TENSORID: Integer number identiying the parameter affected by the fault (0 - No fault, 1 - conv1.w, 2 - conv1.b, 3 - conv2.w, 4 - conv2.b, 5 - fc1.w, 6 - fc1.b, 7 - fc2.w, 8 - fc2.b).</li><li>ELEMID: Integer number identiying the element of the parameter affected by the fault (-1 - No fault, [0-2] - conv1.b, [0-74] - conv1.w, [0-5] - conv2.b, [0-149] - conv2.w, [0-146] - fc1.b, [0-43217] - fc1.w, [0-9] - fc2.b, [0-1469] - fc2.w).</li><li>MASK: 8-digit hexadecimal number identifying those bits affected by the fault ([00000000 - No fault, FFFFFFFF - all 32 bits faulty]).</li><li>FAULT: String identiying the type of fault (NF - No fault, BF - bit-flip, S0 - Stuck-at-0, S1 - Stuck-at-1).</li><li>SOFTMAX: 10 decimal numbers obtained after applying the softmax function to the provided output. They represent the probability of the image of belonging to the corresponding category for classification.</li><li>PRED: Integer number representing the category predicted for the processed image.</li><li>LABEL: integer number representing the actual category for the processed image.</li></ul>

opencc-by-4.0May 2022View details →

Designing Optimal Convolutional Neural Network Architecture Using Differential Evolution Algorithm

Convolutional Neural Networks (CNNs) are widely used deep learning models for solving various tasks such as computer vision, speech recognition, among others. However, CNNs are developed manually based on problem-specific domain knowledge and tricky settings, which are laborious, time-consuming and challenging. To address these issues, this study proposes an Improved Differential Evolution of Convolutional Neural Network algorithm, namely IDECNN, to design CNN layer architectures for image classification task.

dryad36/100

Data from: A convolutional neural network to identify mosquito species (Diptera: Culicidae) of the genus Aedes by wing images

Accurate species identification is a prerequisite to assess the medical relevance of a mosquito specimens. In monitoring or surveillance programs, mosquitoes are typically identified based on morphological characters, which can be supported by molecular biological assays. Both methods require intensive experience of the observers and well-equipped laboratories. The use of convolutional neural networks (CNNs) to identify species based on images may be a cost-effective and reliable alternative. In this proof-of-concept study, we developed a CNN to identify seven Aedes species by wing images, only. While previous studies used images of the whole mosquito body, the nearly two-dimensional wings may facilitate standardized image capture and thereby reduce the complexity of the CNN implementation. Mosquitoes were sampled from different sites in Germany. Their wings were mounted and photographed with a professional stereomicroscope. The data set consisted of 1,155 wing images from seven Aedes species, including the exotic species Aedes albopictus und six native Aedes species, as well as 554 wings from different non-Aedes mosquitoes. The wing images were used to train a CNN to differentiate between Aedes and non-Aedes mosquitoes and to classify the seven Aedes species. The training was conducted separately for grayscale and RGB images. Image processing, data augmentation, training, validation and testing were conducted in python using deep-learning framework PyTorch. For both input images, i.e. grayscale and RGB images, our best-performing CNN configuration achieved an accuracy of 100% to discriminate Aedes from non-Aedes mosquito species. The accuracy to predict the Aedes species reached 93% for grayscale images and 96% for RGB images. Aedes albopictus could be identified with an accuracy of 100%. In conclusion, wing images are sufficient to identify mosquito species by CNN based image classification. Thus, wing images can represent a useful complement for CNN-based image classification, e.g. for damaged mosquito specimens. Larger training data sets with further mosquito species and a greater variety of images are required to improve and test broad applicability.

opencc-zeroFeb 2024View details →

opencc-by-4.0Mar 2024View details →

Rapid and Accurate Identification of Stem Cell Differentiation Stages via SERS and Convolutional Neural Networks

<a name="OLE_LINK2"></a>Monitoring the transition of cell states during induced pluripotent stem cell (iPSC) differentiation is crucial for clinical medicine and basic research. However, both identification category and prediction accuracy need further improvement. Here, we propose a method combining Surface-Enhanced Raman spectroscopy (SERS) with convolutional neural networks (CNN) to precisely identify and distinguish cell states during stem cell differentiation. First, mitochondria-targeted probes were synthesized by combining AuNRs and mitochondrial localization signal (MLS) peptides to obtain effective and stable SERS spectra signals at various stages of cell differentiation. Then, the SERS spectra served as input datasets, and their distinctive features were learned and distinguished by CNN. As a result, rapid and accurate identification of six different cell states, including the embryoid body (EB) stage, was successfully achieved throughout the stem cell differentiation process with an impressive prediction accuracy of 98.5%. Furthermore, the impact of different spectral feature peaks on the identification results was investigated, which provides a valuable reference for selecting appropriate spectral bands to identify cell states. This is also beneficial for shortening the spectral acquisition region to enhance spectral acquisition speed. These results suggest the potential for SERS-CNN models in quality monitoring of stem cells, advancing the practical applications of stem cells.

opencc-by-4.0Nov 2021View details →

Computer-aided Veress needle guidance using endoscopic optical coherence tomography and convolutional neural networks

During laparoscopic surgery, the Veress needle is commonly used in pneumoperitoneum establishment. Precise placement of the Veress needle is still a challenge for the surgeon. In this study, a computer-aided endoscopic optical coherence tomography (OCT) system was developed to effectively and safely guide Veress needle insertion. This endoscopic system was tested by imaging subcutaneous fat, muscle, abdominal space, and the small intestine from swine samples to simulate the surgical process, including the situation with small intestine injury. Each tissue layer was visualized in OCT images with unique features and subsequently used to develop a system for automatic localization of the Veress needle tip by identifying tissue layers (or spaces) and estimating the needle-to-tissue distance. We used convolutional neural networks (CNNs) in automatic tissue classification and distance estimation. The average testing accuracy in tissue classification was 98.53±0.39%, and the average testing relative error in distance estimation reached 4.42±0.56% (36.09±4.92 μm). The dataset is split into two parts: (1) Classification. The zip file veress_classification_raw_images.zip contains 40K images from 8 swine samples where there are 1K images per layer (skin, fat, muscle, abdominal space, and small intestine) (2) Regression. The zip file veress_regression_raw_images.zip contains 8K images of the abdominal space from the same 8 swine samples, and the ground truth distance labels for each sample are found in the Excel files S[1-8]_distance_measurement_20210803.xlsx.

opencc-by-4.0Feb 2022View details →

Convolutional neural network and data used for applied soundscape classification with Soundscapes 2 Landscapes (S2L)

This repository documents the ABGQI-CNN manuscript (DOI: <a href="https://doi.org/10.1016/j.ecolind.2022.108831">https://doi.org/10.1016/j.ecolind.2022.108831</a>). It contains supplementary materials, data used to train a soundscape classification convolutional neural network (CNN), and data to generate manuscript results. The accompanying code can be found at <a href="https://doi.org/10.5281/zenodo.6038460">https://doi.org/10.5281/zenodo.6038459</a>. Files include: <ul> <li>ABGQI-CNN.tar: saved CNN model weights for the 5-class soundscape classifier using a MobileNetV2 architecture pre-trained with bird vocalization data.</li> <li>ABGQI_mel_spectrograms.tar: spectrograms used for fine-tuning the pre-trained CNN, above, with training, validation, and testing data splits.</li> <li>freesound_licensing.csv: file names and license information related to Freesound auxiliary files.</li> <li>RavenLite_Training_Data_Collection.pdf: a manual for RavenLite ROI annotation.</li> <li>S2L_site_geog-env_data.csv: environmental and geographic data (sans GPS) related to site locations in S2L project 2017-2020.</li> <li>site_avg_ABGQIU_fscore_075_daytime.csv: the average site rate of soundscape components for 5 a.m. to 8 p.m.</li> <li>site_by_hour_ABGQIU_fscore_075.csv: the average hourly site rate of soundscape components</li> <li>site_classifications_beta075.tar: a directory containing a CSV for every site with threshold optimized classifications for each 2-s Mel spectrogram</li> <li>site_prediction_probabilies.tar: a directory containing a CSV for every site with ABGQI-CNN probabilities for each 2-s Mel spectrogram</li> <li>Supplementary_Materials.pdf: includes additional material and analyses related to the accompanying manuscript. </li> </ul> Contact Colin Quinn at cq73@nau.edu for questions related to this repository or if you have an interest in the original wav recordings. Please be aware that underlying software, specifically for the CNN implementation, may not continue stability as python libraries are updated.