6 Feb 2023

Three-Dimensional Geospatial Product Generation from Tactical Sources, Co-Registration Assessment, and Considerations

Abstract: According to Army Multi-Domain Operations (MDO) doctrine, generating timely, accurate, and exploitable geospatial products from tactical platforms is a critical capability to meet threats. The US Army Corps of Engineers, Engineer Research and Development Center, Geospatial Research Laboratory (ERDC-GRL) is carrying out 6.2 research to facilitate the creation of three-dimensional (3D) products from tactical sensors to include full-motion video, framing cameras, and sensors integrated on small Unmanned Aerial Systems (sUAS). This report describes an ERDC-GRL processing pipeline comprising custom code, open-source software, and commercial off-the-shelf (COTS) tools to geospatially rectify tactical imagery to authoritative foundation sources. Four datasets from different sensors and locations were processed against National Geospatial-Intelligence Agency–supplied foundation data. Results showed that the co-registration of tactical drone data to reference foundation varied from 0.34 m to 0.75 m, exceeding the accuracy objective of 1 m described in briefings presented to Army Futures Command (AFC) and the Assistant Security of the Army for Acquisition, Logistics and Technology (ASA(ALT)). A discussion summarizes the results, describes steps to address processing gaps, and considers future efforts to optimize the pipeline for generation of geospatial data for specific end-user devices and tactical applications.

23 Apr 2020

PUBLICATION NOTICE: Use of Convolutional Neural Networks for Semantic Image Segmentation Across Different Computing Systems

ABSTRACT: The advent of powerful computing platforms coupled with deep learning architectures have resulted in novel approaches to tackle many traditional computer vision problems in order to automate the interpretation of large and complex geospatial data. Such tasks are particularly important as data are widely available and UAS are increasingly being used. This document presents a workflow that leverages the use of CNNs and GPUs to automate pixel-wise segmentation of UAS imagery for faster image processing. GPU-based computing and parallelization is explored on multi-core GPUs to reduce development time, mitigate the need for extensive model training, and facilitate exploitation of mission critical information. VGG-16 model training times are compared among different systems (single, virtual, multi-GPUs) to investigate each platform’s capabilities. CNN results show a precision accuracy of 88% when applied to ground truth data. Coupling the VGG-16 model with GPU-accelerated processing and parallelizing across multiple GPUs decreases model training time while preserving accuracy. This signifies that GPU memory and cores available within a system are critical components in terms of preprocessing and processing speed. This workflow can be leveraged for future segmentation efforts, serve as a baseline to benchmark future CNN, and efficiently support critical image processing tasks for the Military.