Search NASA⌕ Search

SEARCH · Search NASA

Results for “Labeled Datasets”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 73 records · Page 4

Revolutionizing Earth Science with Generalized AI Models

Foundation Models (FM) are generalized Artificial Intelligence (AI) models that are designed to replace a task or an application-specific model and can be used for many downstream applications. These FM can be built on any sequence data and are trained utilizing self-supervised approaches. The obstacle of creating a sizable labeled dataset for training is removed by using self-supervised learning. Most FM employ transformer design that takes advantage of the idea of self-attention, allowing the network to represent the impact of distant data points on one another in space and time. The FM models show emergent qualities that are induced from the data. FM can become a valuable tool for Earth science researchers. Due to the size of these models, downstream applications built fine-tuning these FM perform better and exhibit greater accuracy than models created from scratch. FM significantly lowers the entry barrier in terms of both the time and effort required to develop various downstream applications. For some scientific datasets, such as optical remote sensing data, FM can speed up processes like classification, object detection and prediction. By eliminating the training data bottleneck and maximizing the usage of science data, FM can make it simpler to integrate AI into scientific research. Initial results for three different FMs will be presented.

Rahul Ramachandran↗

Anomaly Detection for the Roman Space Telescope Wide Field Instrument’s Science Data Processing Pipeline

The Roman Space Telescope (RST) Wide Field Instrument (WFI) will be utilizing a preliminary Science Data Processing (SDP) pipeline during its Integration and Test, and to some extent during Operations, to track basic statistics and identify known features such as cosmic rays, snowballs as well as possible anomalies in raw detector data. In our detectors, these anomalies appear as jumps in the ramp of a readout and are classified as cosmic rays if they appear as a streak or snowballs if they’re more circular. The WFI employs an array of 18 H4RG-10 detectors that collect image samples. Each set of raw frames within a non-destructive exposure is packaged by the SDP pipeline into image cubes for each detector. Each cube is a time series of 4096 × 4096 accumulating pixel frames. The preliminary analysis pipeline is used to locate anomalies in these time-series accumulation frames and identify the type of anomaly, either natural phenomena or detector characteristic. To compare different methods, we’ve implemented both heuristic-based and data-driven methods to identify anomalies. For the heuristic-based approach, we identify snowballs and cosmic rays by the size and shape of outlier pixel clusters between consecutive frames. For data driven methods, we evaluated a Convolutional Neural Network (CNN) model, and more traditional methods like Principal Component Analysis (PCA). CNN is a supervised learning/classification method. Thus, we used a labeled dataset of anomalies to perform segmentation of the image and identify anomalies. We used previously identified cosmic rays and snowballs to measure the accuracy and efficiency of the mentioned approaches. In evaluating these methods, we aim to pick the best fit for the SDP pipeline’s anomaly detection in terms of both performance and runtime.

Paul Horton↗

The Value of Energy Performance and Green Attributes in Buildings: A Review of Existing Literature and Recommendations for Future Research

Labels, certifications, and rating systems for energy efficiency performance and “green” attributes of buildings have been available in the U.S. for over 10 years, and used extensively in the European Union and Australia for longer. Such certifications and ratings can make energy efficiency more visible, and could help spur demand for energy efficiency if these designations are shown to have a positive impact on sales or rental prices. This policy brief discusses the findings and methodologies from recent studies on this topic, and suggests recommendations for future research. Although there have been just a handful of studies within the last 10 years that have investigated these effects, a few key findings emerge: To maximize sales price impact, label or rating information must be disclosed early and visibly in the sales process; The approach to evaluating energy efficiency labels (e.g., ENERGY STAR) and general “green” certifications (e.g., LEED or GreenPoint Rated) may need to be different, depending on the type, vintage and market penetration of the label; Collaborative efforts to promote label adoption and build a large dataset of labeled buildings will be required to produce reliable study results.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

Multiperspective Automotive Labeling

Supervised machine learning techniques inherently rely on datasets to be trained. With image datasets traditionally being annotated by humans, many advancements in image annotation tools have been made to ensure creation of rich datasets with accurate labels. Nevertheless, users still find it challenging to create and use their own datasets with labels that reflect their problem domain. We propose a streamlined labeling process that aligns multiperspective images and allows a transition from a labeled perspective to other perspectives. The main goal of this work is to reduce the human effort required for labeling vehicle images under favorable conditions where the image perspectives are correlated and one or more perspectives are known. A case study is described and analyzed to show the effectiveness of the process, as well as constraints and limitations when applied to other cases.

training dataset, multiperspective images, supervi↗

Image Labeler: A Web Interface to Catalog Earth Science Events

Advances in machine learning (ML) have made it possible to automatically detect Earth science phenomena from satellite imagery. While useful, ML algorithms typically require an extensive dataset containing labeled images for training. Systematic labeling and management of such datasets is quite cumbersome. With this in mind, we present the Image Labeler. Image Labeler is a fast and scalable cloud-based tool that facilitates the rapid development of Earth science event databases, in order to aid automated ML-based image classification.

Case Study↗

Dataset: Occupancy Detection, Tracking, and Estimation Using a Vertically Mounted Depth Sensor

Occupancy detection, tracking, and estimation has a wide range of applications including improving building energy efficiency, safety, and security of the occupants. As depth sensors are getting cheaper, they offer a viable solution to estimate occupancy accurately in a non-privacy invasive manner. Even though there are publicly available depth datasets, they do not consider placing the sensor in the ceiling looking downwards to estimate occupancy. We deployed four Kinect for XBOX One in four CMU classrooms and conference rooms for a period of four weeks in 2017 and collected over 6 TB of depth data. We annotate this huge dataset by labelling bounding boxes around occupants and release the annotated dataset.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Dataset: Occupancy Detection, Tracking, and Estimation Using a Vertically Mounted Depth Sensor

Occupancy detection, tracking, and estimation has a wide range of applications including improving building energy efficiency, safety, and security of the occupants. As depth sensors are getting cheaper, they offer a viable solution to estimate occupancy accurately in a non-privacy invasive manner. Even though there are publicly available depth datasets, they do not consider placing the sensor in the ceiling looking downwards to estimate occupancy. We deployed four Kinect for XBOX One in four CMU classrooms and conference rooms for a period of four weeks in 2017 and collected over 6 TB of depth data. We annotate this huge dataset by labelling bounding boxes around occupants and release the annotated dataset.

47 OTHER INSTRUMENTATION↗

Enhancing Electron Microscopy Image Classification Using Data Augmentation

Manual labeling for machine learning tasks such as image classification is tedious and labor-intensive; as a result, scientific datasets suitable for deep learning applications are scarce and limited. While data augmentation techniques have shown promise for extending image datasets, very little work has been done to understand the impact of combining multiple augmentation methods sequentially or the limits of their effectiveness when combined. Our work addresses this gap by examining how standard and combinatorial data augmentation affects the performance of machine learning models when trained on small datasets for label classification tasks. For our analysis, we generate single, double and quadruple-augmented datasets for a microscopy image classification task using six standard augmentation methods, and compare the resultant improvements observed in binary classification accuracy with three standard image classification models (DenseNet169, MobileNetV2, ResNet101V2). Our experiments show a non-monotonic relationship between the number of simultaneous augmentation methods and classification accuracy, indicating that there is a trade-off between the degree of augmentation and the model performance. These findings suggest that the optimal number of augmentation methods will vary by domain and use case. We also find that the order in which augmentation methods are applied to a limited dataset matters when combining augmentation schemes, with our use case showing performance differences up to 2.6% when the augmentation order is reversed for double-augmented datasets. Our work offers insights to the limits of data augmentation when working on image classification tasks with limited datasets.

Welsman, Jordan A↗

ODI notebook additive_manufacturing_video_2022

Additive Manufacturing, video dataset Two-photon lithography (TPL) is a widely used 3D nanoprinting technique that uses laser light to create objects. Challenges to large-scale adoption of this additive manufacturing method include identifying light dosage parameters and monitoring during fabrication. A research team from LLNL, Iowa State University, and Georgia Tech is applying machine learning models to tackle these challenges-i.e., accelerate the process of identifying optimal light dosage parameters and automate the detection of part quality. Funded by LLNL's Laboratory Directed Research and Development Program, the project team has curated a video dataset of TPL processes for parameters such as light dosages, photo-curable resins, and structures. Both raw and labeled versions of the datasets are available on the links in the Open Data Initiative page. The code uses the labeled dataset. Notebook compiled by Nisha Mulakken (mulakken1@llnl.gov) for LLNL Open Data Initiative, Summer 2022. Original code provided by research team. Publications: X.Y. Lee, S.K. Saha, S. Sarkar, B. Giera. "Automated detection of part quality during two-photon lithography via deep learning." Additive Manufacturing 36, December 2020: doi.org/10.1016/j.addma.2020.101444 X.Y. Lee, S.K. Saha, S. Sarkar, B. Giera. "wo Photon lithography additive manufacturing: Video dataset of parameter sweep of light dosages, photo-curable resins, and structures." Data in Brief 32, October 2020. doi.org/10.1016/j.dib.2020.106119.

Mulakken, NishaJ↗

Creating a Training Dataset for Semantic Segmentation of Canal Networks for Irrigation Modernization

Canal infrastructure has provided critical irrigation water to the western United States for over a century. To continue providing vital water resources to the semi-arid West, irrigation systems must undergo maintenance and modernization. Many canal companies are resource-constrained, and because funding opportunities often require detailed knowledge of existing infrastructure, they can struggle to secure financial capital. We address this problem by creating training data for a semantic segmentation deep learning model to map canal networks throughout the western United States. To create a diverse and robust training dataset, we labelled 1-m NAIP imagery with the locations of no canals, wet canals, and dry/vegetated canals. Since creating these datasets is time consuming, we first developed a preprocessing methodology to identify canals within our four study areas. We used NAIP imagery and provided canal centerline data to buffer, standardize, and cluster the imagery, automating the labeling process as much as possible. However, this still required manual cleaning and manual classification of canal type. Challenges arose when canals were interrupted (e.g., road culverts or piped sections) or when nearby features shared similar characteristics (e.g., irrigated fields, trees, and shadows). Combining automated preprocessing with manual refinement produced four detailed canal masks to be used in the semantic segmentation model developed by Richard Tapia.

13 - HYDRO ENERGY↗

Active Learning for Language Modeling

Foreign disinformation campaigns undermine national security. Various supervised language modeling techniques in NLP can help to understand and dismantle these campaigns, but they rely heavily on large, labeled (often by humans) datasets. This work provides a solution to this problem in the form of an active learning (AL) framework, which is used to generate labeled datasets and leverage human input for detecting disinformation. The developed AL framework utilizes task adaptive pretraining to fully leverage the unlabeled data and boost the performance of the classifier used for labeling. A disinformation rhetoric metric was developed to measure the presence of common rhetorical techniques used in text that are meant to deceive, for both the classifier and human to use in the task of identifying disinformation. This metric was combined with an uncertainty criterion to create a hybrid acquisition method for AL, and this hybrid method was tested alongside other acquisition functions. A sophisticated and robust stopping strategy was developed to signal the AL process should terminate, saving human time from being wasted on iterations that would not significantly benefit classifier performance.

99 GENERAL AND MISCELLANEOUS↗

AMVOS: Additive Manufacturing Video Object Segmentation Dataset

This dataset provides labeled video frames from four additive manufacturing (AM) processes for video object segmentation (VOS) tasks. It contains 90 video segments comprising 900 individually annotated frames across five AM datasets: laser hot-wire directed energy deposition (LHW-DED), tungsten inert gas wire arc additive manufacturing (TIG-WAAM), plasma arc welding (PAW), visible-light polymer extrusion (visPolymer), and near-infrared polymer extrusion (irPolymer). Each video segment consists of 10 contiguous frames with corresponding pixel-level object instance annotations. Depending on the process, two of four object classes are labeled per frame: Melt Pool, Feed Wire, Nozzle, or Material. Raw frames are provided as .jpg files and annotations as palettized .png files. The dataset follows the directory structure of established VOS benchmarks (DAVIS, YouTube-VOS, MOSE), enabling direct integration into VOS model training and evaluation pipelines for foundation model fine-tuning, domain adaptation, or zero-shot performance benchmarking. Data was collected at Oak Ridge National Laboratory's Manufacturing Demonstration Facility.

Wetzel, Jon [ORNL]↗

Validation of RELAP-7 for Forced Convective and Natural Circulation Reactor Flows

New reactor designs such as small modular reactors and newly proposed passive safety systems rely on natural circulation two-phase flow cooling of the reactor core under accident conditions. Without a pump to drive the flow, the circulation rate is a strong function of void fraction. Under these conditions where the flow rate, pressure drop, and heat transfer are all closely coupled through the mixture density, the accurate prediction of void fraction is pivotal. Furthermore, while these passive systems boast a more reliable cooling option, the sensitivity of the system to two-phase flow instabilities is more likely than in forced convective cooling and can result in dangerous density wave or flow excursion conditions. Improvement of one-dimensional system codes and further development of three-dimensional code capability are all important efforts in addressing these issues. The proposed work is a critical step towards providing confidence through validation of the capability of thermal-hydraulic tools to meet the emerging challenges in reactor simulation. Historically, system analysis codes develop a stiffness over time due to fine-tuning of adjustable code parameters that result in a reluctance or inability to adopt improved models. Therefore, a rigorous effort should be invested in preparation for RELAP-7 to identify required modeling improvements of constitutive relations and validation of the code applicability to the wide range of possible reactor flow conditions. Part of this validation involves the uncertainty quantification which is a critical step in the Code Scaling, Applicability, and Uncertainty (CSAU) methodology and Best-Estimate Plus Uncertainty (BEPU) licensing method. BEPU allows recovery of safety margins and potentially optimizing plant performance, resulting in economic benefit for the reactor operators. Beginning with the collection of past experimental data covering prototypic LWR conditions, this document reviews three critical datasets available for the validation of system codes. The experimental methods, procedures, facilities, instrumentation, and supplementary information are consolidated to aid the validation efforts of this work and future projects. These three datasets (824 conditions) are chosen due to their wide acceptance and demonstrated application for code validation. Two experimental datasets from the Nuclear Power Engineering Corporation (NUPEC) span prototypic conditions of Boiling Water Reactors (BWR) and Pressurized Water Reactors (PWR). These datasets are labeled as the BFBT (BWR Full-size Fine-mesh Bundle Test) and the PSBT (PWR Subchannel and Bundle Test) datasets respectively. The third dataset is taken in an internally heated annulus that is scaled based on a BWR subassembly and includes two-phase flows important to post-blowdown analysis. Together, the datasets consider the full range of pressure and mass flux expected in LWR forced convective flows from steady-state to long term cooling conditions. Recognizing the deficiency of available data under low-pressure, low-flow, a new natural circulation dataset consisting of 107 conditions is obtained. With system pressures ranging between 145 and 950 kPa, natural circulation is driven by a heat flux up to 275 kW/m2 in a 3 m long heated region of the annulus test section which is followed by a 2 m long unheated region before entering a condenser. Stable natural circulation flow rates in this facility were measured to be between 182 and 590 kg/m2-s. Pressure and temperature are measured at ten locations in the facility. In addition, measurements of void fraction, interfacial area concentration, and gas velocity are obtained at five axial locations in the annulus test section. The RELAP5-3D code demonstrates good accuracy for void fraction predictions in BFBT and PSBT benchmarks, but shows relatively large discrepancy in the annulus experiments at lower pressures. In two-phase natural circulation, large oscillations are observed in many system parameters during the simulation of 19 experiment cases which were conducted at low pressures, and the model shows good predictive performance with the remaining 88 experiment cases. However, some discrepancies in void fraction and temperature predictions are found which may require improved constitutive modeling under these low-pressure, low-flow condition. The quantified uncertainty in void fraction predictions in three forced convection benchmarks due to uncertainties in boundary conditions and geometries are relatively small, except for a few cases in the annulus experiment. The output uncertainty of the natural circulation model is quantified by considering the measurement error in boundary conditions, and it is shown that the output uncertainty ranges are relatively small for the range of conditions considered. Through this work, a clear pathway is provided for validation of RELAP7 once the code is available.

42 ENGINEERING↗

Augmented Reality Data Generation for Training Deep Learning Neural Network

One of the major challenges in deep learning is retrieving sufficiently large labeled training datasets, which can become expensive and time consuming to collect. A unique approach to training segmentation is to use Deep Neural Network (DNN) models with a minimal amount of initial labeled training samples. The procedure involves creating synthetic data and using image registration to calculate affine transformations to apply to the synthetic data. The method takes a small dataset and generates a highquality augmented reality synthetic dataset with strong variance while maintaining consistency with real cases. Results illustrate segmentation improvements in various target features and increased average target confidence.

Torres, Gil↗

AssistTaxi: A Comprehensive Dataset for Taxiway Analysis and Autonomous Operations

The availability of high-quality datasets play a crucial role in advancing research and development especially, for safety critical and autonomous systems. This poster presents AssistTaxi, which is a comprehensive novel dataset which is a collection of images for runway and taxiway analysis. The dataset comprises of more than 300,000 frames of diverse and carefully collected data, gathered from Melbourne (MLB) and Grant-Valkaria (X59) general aviation airports. The importance of AssistTaxi lies in its potential to advance autonomous operations, enabling researchers and developers to train and evaluate algorithms for efficient and safe taxiing. Researchers can utilize AssistTaxi to benchmark their algorithms, assess performance, and explore novel approaches for runway and taxiway analysis. Additionally, the dataset serves as a valuable resource for validating and enhancing existing algorithms as well as facilitating innovation in autonomous operations for aviation. We also propose an initial approach to label the dataset using a contour based detection and line extraction technique.

Data Collection↗

AI Foundation Models for Science: An Open Collaborative Initiative

Foundation Models (FMs), AI models designed to replace task-specific models, are increasingly being recognized for their versatility across numerous downstream applications. These models, trained using self-supervised techniques on any type of sequence data, circumvent the need for large annotated datasets, a major bottleneck in traditional AI model development. FMs can be applied to downstream tasks using few-shot learning and fine-tuning, significantly reducing the need for large labeled training datasets and computational resources. However, the development of FMs requires substantial resources, including access to data and compute power, expertise in the latest models, and specialized scientific knowledge for systematic evaluation. It is challenging for a single group to possess all these capabilities. To address this, NASA IMPACT has initiated an open collaborative effort, leveraging partnerships with the private sector and other groups within and outside NASA, to jointly build FMs. The overarching goal is to develop a consistent and collaborative approach to building FMs for high-value science datasets. This initiative has fostered collaboration within NASA and with external partners, including IBM Research, Clark University, DOE’s ORNL, ESA, and USGS. The effort focuses on identifying key datasets with a wide range of downstream applications, pretraining and building FMs using modified transformer architectures, evaluating compute infrastructure needs, and sharing models, pretraining and fine-tuning code, and data with the community. Furthermore, it aims to train the Earth science community to fine-tune these models for various downstream applications. Our initial effort resulted in the creation of a 100 million parameter HLS Geospatial Model within six months, which was released on HuggingFace. We are now expanding our scope to include data from weather and climate models and investigating multimodal models. We invite those interested in participating in this effort to join us by sharing their use cases, expertise, or data.

Rahul Ramachandran↗

Describing Point Defect Topology in 2D Energy Materials Through Computer Vision

Point defects such as vacancies and impurity atoms strongly impact the performance of 2D materials. Traditional efforts often rely on manual detection, a process that is time-intensive, prone to human error, and challenging to scale. Here we leverage machine learning (ML) methods to identify and quantify vacancies within 2D transition metal carbides (Ti3C2, MXenes), aiming to expedite detection while improving accuracy. MXenes exhibit valuable defect-defined electrochemical properties, but we currently lack statistical understanding of defect topology needed to fully harness these materials. Here we employ a convolutional neural network for semantic segmentation of experimental MXene images, opening an opportunity to conduct a rigorous statistical study on defect hierarchy while investigating local relaxation in the lattice. We show how the integration of ML can yield fundamental insight into point defects, providing a powerful tool that will play an increasingly crucial role in the future of materials science. ML is often not just a matter of straightforward application, and pretrained models proved ineffective in this case. Instead, we trained our own neural network (NN) and applied data augmentation techniques and fine-tuning to the training dataset. Since labeled microscopy data is often scarce, we developed training data from a previously published wide-frame MXene image, using customized Gaussian fitting to locate atomic positions. Our trained model was then applied to a large dataset of experimental images, enabling a statistical study of defect configurations across three samples prepared with different HF etchant concentrations (5%, 9.1%, and 12.5%), as shown in Fig. 1. This also allowed us to investigate local strain around vacancies, though we find that we are limited by the precision of measurements using high-angle annular dark field (HAADF) images, as shown in Fig. 2. This study demonstrates how ML enables large-scale, quantitative analysis of atomic defects - an otherwise infeasible task with traditional methods. While our NN was specialized for Ti3C2 MXenes, the pipeline we developed provides a foundation for future ML models tailored to other materials. Ultimately, we envision embedding the NN onto the microscope to give real-time feedback to the user. To make this a reality, continued work is necessary to fully understand the NN's capabilities and limitations. This study gets one step closer to our goals of automated experimentation moving away from traditional methods of manual labeling. As ML capabilities advance, we hope to continue adapting and applying these techniques in microscopy.

2D materials↗

Statistical and Neural Network for Real Sensor-Data-Driven Anomaly Detection in Nuclear Applications

Anomaly detection (AD) in sensor data is critical to ensure uninterrupted functionality of nuclear power plants (NPPs). Consequently, AD model validation through real-world sensor data is important for applications in nuclear facilities. In this paper, we propose an Autoencoder (AE)—a multi-layered neural network, for AD in sensor data from an operational NPP testbed. Since the dataset lacks labels for irregularities, we introduce random noise and label them to effectively train our model. The proposed AE model assigns a higher reconstruction error to the abnormal samples that deviate from those encountered during the training phase and uses the reconstruction loss to detect anomalies in a representative imbalanced dataset. We also introduce an analytical solution—seasonal trend decomposition (STD)—as another AD scheme for identifying irregularities withinthe same time-series dataset. In contrast to the AE model which relies on reconstruction loss, the STD scheme decomposes the entire dataset into its trend, seasonality, and residual components to pinpoint irregularities. Our findings indicate that the proposed AE and STD models individually achieve recall scores of 97% and 92%, respectively. We validate the performance of the two models on both balanced and imbalanced data. We further solidify the results by picking the combined selected anomalies of the two solutions with an "AND" operator for more reliable predictions.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗