skip to main content

Explore Research Products in the NSF-PAR

Title: Knowledge-Driven GeoAI: Integrating Spatial Knowledge into Multi-Scale Deep Learning for Mars Crater Detection
This paper introduces a new GeoAI solution to support automated mapping of global craters on the Mars surface. Traditional crater detection algorithms suffer from the limitation of working only in a semiautomated or multi-stage manner, and most were developed to handle a specific dataset in a small subarea of Mars’ surface, hindering their transferability for global crater detection. As an alternative, we propose a GeoAI solution based on deep learning to tackle this problem effectively. Three innovative features are integrated into our object detection pipeline: (1) a feature pyramid network is leveraged to generate feature maps with rich semantics across multiple object scales; (2) prior geospatial knowledge based on the Hough transform is integrated to enable more accurate localization of potential craters; and (3) a scale-aware classifier is adopted to increase the prediction accuracy of both large and small crater instances. The results show that the proposed strategies bring a significant increase in crater detection performance than the popular Faster R-CNN model. The integration of geospatial domain knowledge into the data-driven analytics moves GeoAI research up to the next level to enable knowledge-driven GeoAI. This research can be applied to a wide variety of object detection and image analysis tasks.  more » « less
Award ID(s):
1853864 1455349
NSF-PAR ID:
10276644
Author(s) / Creator(s):
; ;
Date Published:
Journal Name:
Remote Sensing
Volume:
13
Issue:
11
ISSN:
2072-4292
Page Range / eLocation ID:
2116
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ; ; ; ; ; ; ; ; et al ( , Astronomy & Astrophysics)
    null (Ed.)
    Context. Asteroid (7) Iris is an ideal target for disk-resolved imaging owing to its brightness ( V ~ 7–8) and large angular size of 0.33′′ during its apparitions. Iris is believed to belong to the category of large unfragmented asteroids that avoided internal differentiation, implying that its current shape and topography may record the first few 100 Myr of the solar system’s collisional evolution. Aims. We recovered information about the shape and surface topography of Iris from disk-resolved VLT/SPHERE/ZIMPOL images acquired in the frame of our ESO large program. Methods. We used the All-Data Asteroid Modeling ( ADAM ) shape reconstruction algorithm to model the 3D shape of Iris, using optical disk-integrated data and disk-resolved images from SPHERE and earlier AO systems as inputs. We analyzed the SPHERE images and our model to infer the asteroid’s global shape and the morphology of its main craters. Results. We present the 3D shape, volume-equivalent diameter D eq = 214 ± 5 km, and bulk density ρ = 2.7 ± 0.3 g cm −3 of Iris. Its shape appears to be consistent with that of an oblate spheroid with a large equatorial excavation. We identified eight putative surface features 20–40 km in diameter detected at several epochs, which we interpret as impact craters, and several additional crater candidates. Craters on Iris have depth-to-diameter ratios that are similar to those of analogous 10 km craters on Vesta. Conclusions. The bulk density of Iris is consistent with that of its meteoritic analog based on spectroscopic observations, namely LL ordinary chondrites. Considering the absence of a collisional family related to Iris and the number of large craters on its surface, we suggest that its equatorial depression may be the remnant of an ancient (at least 3 Gyr) impact. Iris’s shape further opens the possibility that large planetesimals formed as almost perfect oblate spheroids. Finally, we attribute the difference in crater morphology between Iris and Vesta to their different surface gravities, and the absence of a substantial impact-induced regolith on Iris. 
    more » « less
  2. ; ; ; ; ; ; ; ; ; ( , PLOS ONE)
    Romanach, Stephanie S. (Ed.)
    Massive biological databases of species occurrences, or georeferenced locations where a species has been observed, are essential inputs for modeling present and future species distributions. Location accuracy is often assessed by determining whether the observation geocoordinates fall within the boundaries of the declared political divisions. This otherwise simple validation is complicated by the difficulty of matching political division names to the correct geospatial object. Spelling errors, abbreviations, alternative codes, and synonyms in multiple languages present daunting name disambiguation challenges. The inability to resolve political division names reduces usable data, and analysis of erroneous observations can lead to flawed results. Here, we present the Geographic Name Resolution Service (GNRS), an application for correcting, standardizing, and indexing world political division names. The GNRS resolves political division names against a reference database that combines names and codes from GeoNames with geospatial object identifiers from the Global Administrative Areas Database (GADM). In a trial resolution of political division names extracted from >270 million species occurrences, only 1.9%, representing just 6% of occurrences, matched exactly to GADM political divisions in their original form. The GNRS was able to resolve, completely or in part, 92% of the remaining 378,568 political division names, or 86% of the full biodiversity occurrence dataset. In assessing geocoordinate accuracy for >239 million species occurrences, resolution of political divisions by the GNRS enabled the detection of an order of magnitude more errors and an order of magnitude more error-free occurrences. By providing a novel solution to a significant data quality impediment, the GNRS liberates a tremendous amount of biodiversity data for quantitative biodiversity research. The GNRS runs as a web service and is accessible via an API, an R package, and a web-based graphical user interface. Its modular architecture is easily integrated into existing data validation workflows. 
    more » « less
  3. ; ; ( , Geosciences)
    Globally, coastal zones, rivers and riverine areas, and deltas carry enormous values for ecosystems, socio-economic, and environmental perspectives. These often highly populated areas are generally significantly different from interior hinterlands in terms of population density, economic activities, and geophysical and ecological processes. Geospatial technologies are widely used by scholars from multiple disciplines to understand the dynamic nature of shoreline changes globally. In this paper, we conduct a systematic literature review to identify and interpret research patterns and themes related to shoreline change detection from 2000 to 2021. Two databases, Web of Science and Scopus, were used to identify articles that investigate shoreline change analysis using geospatial technique such as remote sensing and GIS analysis capabilities (e.g., the Digital Shoreline Analysis System (DSAS). Between the years 2000 and 2021, we initially found 1622 articles, which were inspected for suitability, leading to a final set of 905 articles for bibliometric analysis. For systematic analysis, we used Rayyan—a web-based platform used for screening literature. For bibliometric network analysis, we used the CiteSpace, Rayyan, and VOSviewer software. The findings of this study indicate that the majority of the literature originated in the USA, followed by India. Given the importance of protecting the communities living in the riverine areas, coastal zones, and delta regions, it is necessary to ask new research questions and apply cutting-edge tools and technology, such as machine learning approach and GeoAI, to fill the research gaps on shoreline change analysis. Such approaches could include, but are not limited to, centimeter level accuracy with high-resolution satellite imagery, the use of unmanned aerial vehicles (UAV), and point cloud data for both local and global level shoreline change and analysis. 
    more » « less
  4. ; ; ( , The Journal of the Astronautical Sciences)
    Abstract

    It is often necessary to identify a pattern of observed craters in a single image of the lunar surface and without any prior knowledge of the camera’s location. This so-called “lost-in-space” crater identification problem is common in both crater-based terrain relative navigation (TRN) and in automatic registration of scientific imagery. Past work on crater identification has largely been based on heuristic schemes, with poor performance outside of a narrowly defined operating regime (e.g., nadir pointing images, small search areas). This work provides the first mathematically rigorous treatment of the general crater identification problem. It is shown when it is (and when it is not) possible to recognize a pattern of elliptical crater rims in an image formed by perspective projection. For the cases when it is possible to recognize a pattern, descriptors are developed using invariant theory that provably capture all of the viewpoint invariant information. These descriptors may be pre-computed for known crater patterns and placed in a searchable index for fast recognition. New techniques are also developed for computing pose from crater rim observations and for evaluating crater rim correspondences. These techniques are demonstrated on both synthetic and real images.

     
    more » « less
  5. ; ; ( , Transactions in GIS)
    Abstract

    Knowledge graphs are a key technique for linking and integrating cross‐domain data, concepts, tools, and knowledge to enable data‐driven analytics. As much of the world's data have become massive in size, visualizing graph entities and their interrelationships intuitively and interactively has become a crucial task for ingesting and better utilizing graph content to support semantic reasoning, discovering hidden knowledge discovering, and better scientific understanding of geophysical and social phenomena. Despite the fact that many such phenomena (e.g., disasters) have clear spatial footprints and geographic properties, their location information is considered only as a textual label in existing graph visualization tools, limiting their capability to reveal the geospatial distribution patterns of the graph nodes. In addition, most graph visualization techniques rely on 2D graph visualization, which constrains the dimensions of information that can be presented and lacks support for graph structure examination from multiple angles. To tackle the above challenges, we developed a novel 3D map‐based graph visualization algorithm to enable interactive exploration of graph content and patterns in a spatially explicit manner. The algorithm extends a 3D force directed graph by integrating a web map, an additional geolocational force, and a force balancing variable that allows for the dynamic adjustment of the 3D graph structure and layout. This mechanism helps create a balanced graph view between the semantic forces among the graph nodes and the attractive force from a geolocation to a graph node. Our solution offers a new perspective in visualizing and understanding spatial entities and events in a knowledge graph.

     
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email