SEARCH · Search NASA

Results for “scene understanding”

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.

SourceResearch connectionFrom yearThrough year

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

At least 19 records

All sources NASA NTRS DOE OSTI

DOE OSTI2023

Towards Geospatial Knowledge Graph Infused Neuro-Symbolic AI for Remote Sensing Scene Understanding

Deep learning has proven its effectiveness in numerous tasks for remote sensing scene understanding. However there is an increasing interest to explore fusion of domain-specific background information to the deep neural network to further improve its performance. Remote sensing researchers are also working towards developing models that generalize and adapt to multiple applications. Generalization challenges coupled with the scarcity of large corpora of high-quality noise-free labelled data, have together fueled an interest for leveraging background information. Knowledge graphs serve as excellent choice to represent domain-specific information in a structured, standardized and extensible manner. Integrating symbolic knowledge representations in the form of Knowledge Graph Embedding (KGE) to perform neuro-symbolic reasoning is an emerging research direction promising significant impacts. This vision paper seeks to position ideas and provoke early thoughts toward advancing neuro-symbolic artificial intelligence in the context of geospatial challenges. Specifically, it conceptualizes and elaborates on an architecture for infusing geospatial knowledge from knowledge graph in a deep neural network pipeline. As guiding case studies - land-use land-cover classification, object detection and instance segmentation can benefit from infusing spatio-contextual information with remote sensing imagery. The discussion further reflects on and articulates the challenges and explainable AI opportunities anticipated when scaling and maintaining large-scale geospatial knowledge graphs.

Potnis, Abhishek↗

DOE OSTI2024

So you think you can track?

This work introduces a multi-camera tracking dataset consisting of 234 hours of video data recorded concurrently from 234 overlapping HD cameras covering a 4.2 mile stretch of 8-10 lane interstate highway near Nashville, TN. Video is recorded in cooperation with Tennessee State Department of Transportation and its policies. The video is recorded during a period of high traffic density with 500+ objects typically visible within the scene and typical object longevities of 3-15 minutes. GPS trajectories from 270 vehicle passes through the scene are manually corrected in the video data to provide a set of ground-truth trajectories for recall-oriented tracking metrics, and object detections are provided for each camera in the scene (159 million total before cross-camera fusion). Initial benchmarking of tracking-by-detection algorithms is performed against the GPS trajectories, and a best HOTA of only 9.5% is obtained (best recall 75.9% at IOU 0.1, 47.9 average IDs per ground truth object), indicating the benchmarked trackers do not perform sufficiently well at the long temporal and spatial durations required for traffic scene understanding. Video data, scene information, and vehicle trajectories are made publicly available at i24motion.org.

42 ENGINEERING↗

DOE OSTI2021

2021 Smoky Mountains Conference Data Challenge Synthetic-to-Real Domain Adaptation for Autonomous Driving Dataset

The dataset is comprised of both real and synthetic images from a vehicle's forward-facing camera. Each camera image is accompanied by a corresponding pixel-level semantic segmentation image (all files are .png files). In total, the dataset contains 5600 images in the training/validation set and 1400 images in the testing set. The training dataset contains mostly synthetic RGB images collected with a wide range of weather and lighting conditions using the CARLA simulator [1]. In addition, the training data also includes a small pre-selected subset of data from the Cityscapes training dataset â€“ which is comprised of RGB-segmentation image pairs from driving scenarios in various European cities [2]. The testing data is split into three sets. The first set contains synthetic CARLA images with weather/lighting conditions that were not present in the training set. The second set is a subset of the Cityscapes testing dataset. Finally, the third set is an unknown testing set which will not be revealed to the participants until after the submission deadline. [1] Dosovitskiy, A., Ros, G., Codevilla, F., Lopez, A., and Koltun, V. (2017, October). CARLA: An open urban driving simulator. In Conference on robot learning (pp. 1-16). PMLR. [2] Cordts, M., Omran, M., Ramos, S., Rehfeld, T., Enzweiler, M., Benenson, R., ... and Schiele, B. (2016). The cityscapes dataset for semantic urban scene understanding. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 3213-3223).

99 GENERAL AND MISCELLANEOUS↗

DOE OSTI2022

Tackling the Challenges in Scene Graph Generation With Local-to-Global Interactions

In this work, we seek new insights into the underlying challenges of the scene graph generation (SGG) task. Quantitative and qualitative analysis of the visual genome (VG) dataset implies: 1) ambiguity: even if interobject relationship contains the same object (or predicate), they may not be visually or semantically similar; 2) asymmetry: despite the nature of the relationship that embodied the direction, it was not well addressed in previous studies; and 3) higher-order contexts: leveraging the identities of certain graph elements can help generate accurate scene graphs. Motivated by the analysis, we design a novel SGG framework, Local-to-global interaction networks (LOGINs). Locally, interactions extract the essence between three instances of subject, object, and background, while baking direction awareness into the network by explicitly constraining the input order of subject and object. Globally, interactions encode the contexts between every graph component (i.e., nodes and edges). Finally, Attract and Repel loss is utilized to fine-tune the distribution of predicate embeddings. By design, our framework enables predicting the scene graph in a bottom-up manner, leveraging the possible complementariness. To quantify how much LOGIN is aware of relational direction, a new diagnostic task called Bidirectional Relationship Classification (BRC) is also proposed. Overall, experimental results demonstrate that LOGIN can successfully distinguish relational direction than existing methods (in BRC task), while showing state-of-the-art results on the VG benchmark (in SGG task).

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2024

Role of depth in optical diffractive neural networks

Free-space all-optical diffractive neural networks have emerged as promising systems for neuromorphic scene classification. Understanding the fundamental properties of these systems is important to establish their ultimate performance. Here we consider the case of diffraction by subwavelength apertures and study the behavior of the system as a function of the number of diffractive layers by employing a co-design modeling approach. We show that adding depth allows the system to achieve high classification accuracies with a reduced number of diffractive features compared to a single layer, but that it does not allow the system to surpass the performance of an optimized single layer. The improvement from depth is found to be limited to the first few layers. These properties originate from the constraints imposed by the physics of light, in particular the weakening electric field with distance from the aperture.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

DOE OSTI2022

A multi-modal scanning system to digitize CBRNE emergency response scenes

A handheld system developed to digitize a contextual understanding of the scene at a chemical, biological, radiological, nuclear and/or explosives (CBRNE) events is described. The system uses LiDAR and cameras to create a colorized 3D model of the environment, which helps domain experts that are supporting responders in the field. To generate the digitized model, a responder scans any suspicious objects and the surroundings by carrying the system through the scene. The scanning system provides a real-time user interface to inform the user about scanning progress and to indicate any areas that may have been missed either by the LiDAR sensors or the cameras. Currently, the collected data are post-processed on a different device, building a colorized triangular mesh of the encountered scene, with the intention of moving this pipeline to the scanner at a later point. The mesh is sufficiently compressed to be sent over a reduced bandwidth connection to a remote analyst. Furthermore, the system tracks fiducial markers attached to diagnostic equipment that is placed around the suspicious object. The resulting tracking information can be transmitted to remote analysts to further facilitate their supporting efforts. The paper will discuss the system's design, software components, the user interface used for scanning a scene, the necessary procedures for calibration of the sensors, and the processing steps of the resulting data. The discussion will close by evaluating the system's performance on 11 scenes.

Salathe, Marco↗

DOE OSTI2022

A multi-modal scanning system to digitize CBRNE emergency response scenes

Prins, Nicholas↗

DOE OSTI2020

Sampling Subjective Polygons for Patch-Based Deep Learning Land-Use Classification in Satellite Images

Model generalization remains a key challenge in the analysis of large amounts of heterogeneous satellite image data. One major limiting factor in developing generalizable models, in the context of supervised learning, is the lack of high quality training datasets. A model's capacity to perform well on new data is often inhibited by imbalance and bias in the data that was used for training. This is especially a problem when using convolutional neural networks to classify urban land-use in satellite images. Notable dataset imbalance issues in this application include land-use type imbalance and image scene imbalance. To begin understanding these dataset imbalance problems in more detail, we develop and test a number of sampling methods for generating training image datasets from subjective training polygons for urban land-use classification. We investigate sampling at different point densities as a means to reduce content repetition and therefore content imbalance and bias in the training image dataset.

Arndt, Jacob↗

DOE OSTI2022

Transience of the Retinal Output Is Determined by a Great Variety of Circuit Elements

Retinal ganglion cells (RGCs) encrypt stimulus features of the visual scene in action potentials and convey them toward higher visual centers in the brain. Although there are many visual features to encode, our recent understanding is that the ~46 different functional subtypes of RGCs in the retina share this task. In this scheme, each RGC subtype establishes a separate, parallel signaling route for a specific visual feature (e.g., contrast, the direction of motion, luminosity), through which information is conveyed. The efficiency of encoding depends on several factors, including signal strength, adaptational levels, and the actual efficacy of the underlying retinal microcircuits. Upon collecting inputs across their respective receptive field, RGCs perform further analysis (e.g., summation, subtraction, weighting) before they generate the final output spike train, which itself is characterized by multiple different features, such as the number of spikes, the inter-spike intervals, response delay, and the rundown time (transience) of the response. These specific kinetic features are essential for target postsynaptic neurons in the brain in order to effectively decode and interpret signals, thereby forming visual perception. We review recent knowledge regarding circuit elements of the mammalian retina that participate in shaping RGC response transience for optimal visual signaling.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2021

Robotic Understanding of Spatial Relationships Using Neural-Logic Learning

Understanding spatial relations of objects is critical in many robotic applications such as grasping, manipulation, and obstacle avoidance. Humans can simply reason object's spatial relations from a glimpse of a scene based on prior knowledge of spatial constraints. The proposed method enables a robot to comprehend spatial relationships among objects from RGB-D data. This paper proposed a neural-logic learning framework to learn and reason spatial relations from raw data by following logic rules on spatial constraints. The neural-logic network consists of three blocks: grounding block, spatial logic block, and inference block. The grounding block extracts high-level features from the raw sensory data. The spatial logic blocks can predicate fundamental spatial relations by training a neural network with spatial constraints. The inference block can infer complex spatial relations based on the predicated fundamental spatial relations. Simulations and robotic experiments evaluated the performance of the proposed method.

Wang, Dali↗

DOE OSTI2024

Self-Supervised Cloud Classification

Abstract Low-level marine clouds play a pivotal role in Earth’s weather and climate through their interactions with radiation, heat and moisture transport, and the hydrological cycle. These interactions depend on a range of dynamical and microphysical processes that result in a broad diversity of cloud types and spatial structures, and a comprehensive understanding of cloud morphology is critical for continued improvement of our atmospheric modeling and prediction capabilities moving forward. Deep learning has recently accelerated our ability to study clouds using satellite remote sensing, and machine learning classifiers have enabled detailed studies of cloud morphology. A major limitation of deep learning approaches to this problem, however, is the large number of hand-labeled samples that are required for training. This work applies a recently developed self-supervised learning scheme to train a deep convolutional neural network (CNN) to map marine cloud imagery to vector embeddings that capture information about mesoscale cloud morphology and can be used for satellite image classification. The model is evaluated against existing cloud classification datasets and several use cases are demonstrated, including training cloud classifiers with very few labeled samples, interrogation of the CNN’s learned internal feature representations, cross-instrument application, and resilience against sensor calibration drift and changing scene brightness. The self-supervised approach learns meaningful internal representations of cloud structures and achieves comparable classification accuracy to supervised deep learning methods without the expense of creating large hand-annotated training datasets. Significance Statement Marine clouds heavily influence Earth’s weather and climate, and improved understanding of marine clouds is required to improve our atmospheric modeling capabilities and physical understanding of the atmosphere. Recently, deep learning has emerged as a powerful research tool that can be used to identify and study specific marine cloud types in the vast number of images collected by Earth-observing satellites. While powerful, these approaches require hand-labeling of training data, which is prohibitively time intensive. This study evaluates a recently developed self-supervised deep learning method that does not require human-labeled training data for processing images of clouds. We show that the trained algorithm performs competitively with algorithms trained on hand-labeled data for image classification tasks. We also discuss potential downstream uses and demonstrate some exciting features of the approach including application to multiple satellite instruments, resilience against changing image brightness, and its learned internal representations of cloud types. The self-supervised technique removes one of the major hurdles for applying deep learning to very large atmospheric datasets.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2022

Detecting damaged buildings using real-time crowdsourced images and transfer learning

After significant earthquakes, we can see images posted on social media platforms by individuals and media agencies owing to the mass usage of smartphones these days. These images can be utilized to provide information about the shaking damage in the earthquake region both to the public and research community, and potentially to guide rescue work. This paper presents an automated way to extract the damaged buildings images after earthquakes from social media platforms such as Twitter and thus identify the particular user posts containing such images. Using transfer learning and ~ 6500 manually labelled images, we trained a deep learning model to recognize images with damaged buildings in the scene. The trained model achieved good performance when tested on newly acquired images of earthquakes at different locations and when ran in near real-time on Twitter feed after the 2020 M7.0 earthquake in Turkey. Furthermore, to better understand how the model makes decisions, we also implemented the Grad-CAM method to visualize the important regions on the images that facilitate the decision.

58 GEOSCIENCES↗

DOE OSTI2022

Flexible and Accessible 4D Subsurface Visualization Using a Web-Based Platform

4D subsurface visualization using a web platform can provide improved communication, education and outreach to non-experts and stakeholders. It allows for the improved understanding of complex relationships and interactions that occur in inaccessible locations of which many experimental testbeds exist in. The use of a web-based visualization tool lessens the cognitive impact further by reducing the need of specialized software. Users can define their own exploration of a 3D or 4D scene, adding and removing data as needed and are able to make informed decisions based on accurate model visualizations, while providing a collaborative tool that can be accessed on any platform using a browser window. The tool is also able to point to real time streams to display up-to-the-second data as an experiment is ongoing. This level of latency can also provide operators with essential information that can direct an experiment’s progress. Additionally, this tool can leverage augmented and virtual reality (AR, VR) capabilities of certain mobile devices and head-mounted displays, providing further engaging visualization possibilities.

Pratt, Martin J.↗

DOE OSTI2024

Development of a Full-Scale Connected U-Net for Reflectivity Inpainting in Spaceborne Radar Blind Zones

CloudSat’s Cloud Profiling Radar is a valuable tool for remotely monitoring high-latitude snowfall, but its ability to observe hydrometeor activity near the Earth’s surface is limited by a radar blind zone caused by ground clutter contamination. This study presents the development of a deeply supervised U-Net-style convolutional neural network to predict cold season reflectivity profiles within the blind zone at two Arctic locations. The network learns to predict the presence and intensity of near-surface hydrometeors by coupling latent features encoded in blind zone-aloft clouds with additional context from collocated atmospheric state variables (i.e., temperature, specific humidity, and wind speed). Results show that the U-Net predictions outperform traditional linear extrapolation methods, with low mean absolute error, a 38% higher Sørensen–Dice coefficient, and vertical reflectivity distributions 60% closer to observed values. The U-Net is also able to detect the presence of near-surface cloud with a critical success index (CSI) of 72% and cases of shallow cumuliform snowfall and virga with 18% higher CSI values compared to linear methods. An explainability analysis shows that reflectivity information throughout the scene, especially at cloud edges and at the 1.2-km blind zone threshold, along with atmospheric state variables near the tropopause, are the most significant contributors to model skill. This surface-trained generative inpainting technique has the potential to enhance current and future remote sensing precipitation missions by providing a better understanding of the nonlinear relationship between blind zone reflectivity values and the surrounding atmospheric state.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2025

Relative Contributions of sc-DER, mel-DER, Color Rendition, Chromaticity, and Illuminance to Spatial Brightness Perception

An experiment was conducted to examine spectrally based factors that contribute to the visual perception of interior environments, with a focus on brightness perception. Thirty-two participants evaluated 60 different lighting scenes in a mock office. The lighting spectral power distributions varied systematically in illuminance, chromaticity (CCT and Duv), s cone opic daylight efficacy ratio (sc-DER), melanopic daylight efficacy ratio (mel-DER), and color rendition (R f , R g , and R cs,h1 ). At the operationalized levels of these variables, illuminance had the largest effect on brightness perception. Notably, the second largest effect was due to changes in red chroma (R cs,h1 ). The effect of sc-DER was also statistically significant but was a tertiary effect. The effects of mel-DER, CCT, and Duv were not statistically significant. This large effect of red chroma is consistent with the existing understanding that changes in color perception are often perceived when illuminance changes. In conclusion, with appropriate changes in color rendition and other factors held constant, spatial brightness perception was preserved through a decrease from 500 lux to 250 lux.

42 ENGINEERING↗

DOE OSTI2020

Understanding Bifacial PV Modeling: Raytracing and View Factor Models

As part of the PV Magazine Webinar 'View-factor vs. ray tracing - which bifacial modelling techniques should you use?,' the second part lead by NREL will explore the different rear-irradiance calculation software, which fall into two categories: view-factor and ray-tracing models. View factor models assume isotropic scattering of reflected rays, allowing for calculation of irradiance by integration. Due-diligence software such as PVSyst or SAM use the view-factor model. Ray-tracing models simulate multipath reflection and absorption of individual rays entering a scene. Raytracing software such as bifacial_radiance, which is the only available open-source toolkit, offers the possibility of reproducing complex scenes, including shading or finite-system edge effects. This can be useful to evaluate specific system geometries, shading and obstructions, areas of different albedo and electrical mismatch, all topics of interest to the prediction of bifacial performance. Model agreement for view factors and bifacial_radiance software is improving and has been shown to be between 2% (absolute) when compared with measured results.

14 SOLAR ENERGY↗

DOE OSTI2024

Detailed Characterization of CZT Detector Response for Improved Coded-Aperture Imaging Performance

Gamma-ray imaging is a powerful method for locating and quantifying sources of radiation. The coded-aperture technique demonstrates superior angular resolution in comparison to other methods (e.g., Compton reconstruction). In this method, a mask constructed of highly attenuating material encodes the scene as a shadow pattern on a position-sensitive detector; this pattern can then be used to recreate the origin(s) of incident radiation. This is typically done through convolution of the mask and shadow patterns. Iterative methods which attempt to reconstruct the observed shadow pattern using a weighted combination of simulated patterns may also be employed. In either case, errors in event position reconstruction due to detector imperfections alter the shadow pattern and will therefore degrade system performance and may introduce imaging artifacts. These effects can be mitigated with a detailed understanding of such errors – allowing for the generation of representative simulations that include the errors and/or correction of raw imager data to remove the errors. We present a calibration process for a commercially available cadmium zinc telluride (CZT) gamma imager which provides a comprehensive characterization of the spatial and energy dependence of event reconstruction. By illuminating a mask featuring a regular grid of pinholes with a calibration source, the localized response of the detector can be measured with fine granularity. These local responses are combined to generate a full detector response map which can be used to distort simulations in a manner that is representative of the observed detector data. Details of the calibration procedure and an assessment of the impact of its end products on the performance of iterative imaging methods will be presented.

Ziock, Klaus-Peter↗

DOE OSTI2024

Deep Learning Scene Classification Experiments in Automatic Detection of Slums on Planetscope Imagery

Population growth is increasingly happening in slum settlements of the large urban centers in the Global South. The term "slum" encompasses a wide range of communities, located mostly in underserved areas, and often exhibiting distinct structural and functional informalities with a relatively high concentration of marginalized populations. To address the issues confronting slums for effective planning and development, including the realistic estimation of the resident population, identifying them accurately is fundamental. Given the disagreements over a universal definition, diverse characteristic features, and socio-political limitations, global detection of slums is a veritable challenge. In this paper, we present experiments in slum detection using a scene classification algorithm and 3-meter spatial resolution satellite imagery. We train and evaluate the model for slum detection in Mumbai, India for the year 2023 and test the temporal generalization of the trained model on Mumbai in 2020 and 2018. In addition, we explore the pathways toward geographic generalization to Kolkata and Delhi (India). We discuss several limitations in the workflow and model, situate our findings in the existing literature, and suggest improvements and alternatives. With this, we establish baseline methods and experiments as a first step towards developing an image-based global slum detection framework and algorithm. This work adds to the community discussion on methods, data challenges, and open questions related to the detection of slums globally. With this research, we hope to improve our understanding of human settlements, especially in critical areas, improve population estimates, and help measure progress towards the sustainable development goals.

Arndt, Jacob↗