Award Data

On this effort Toyon Research Corp. and The Pennsylvania State University are developing deep learning-based algorithms for object recognition and new class discovery in look-down infrared (IR) imagery. Our approach involves the development of a hybrid classifier that exploits both transfer learning and semi-supervised paradigms in order to maintain good generalization accuracy, especially when li ...

The National Geospatial-Intelligence Agency (NGA) ingests and analyzes raw imagery from multiple sources to form actionable intelligenceproducts that can be disseminated across the intelligence community (IC). To effectively meet these demands NGA must continue to improveits automated and semi-automated methods for target detection and classification. Of particular concern is furthering NGA's abil ...

Toyon Research Corp. proposes development of a system that automates disaster assessment based on fusion of overhead and ground-basedimages, video, and other data. In Phase I, we will investigate various possible data sources and the benefits of fusing the data in automatedanalysis. We will select and curate data for processing in a Phase I feasibility study. Damage assessment will be performed in ...

Toyon Research Corporation proposes to research and develop algorithms for generalized change detection, by leveraging and exploringexisting and proven effective traditional and deep learning methods, with a unique 3D reconstruction component. The vast majority of themassive amounts of imagery data will have small pixel level differences due to a multitude of unimportant changes: minor misregistra ...

The NGA requires an automatic, unsupervised SAR feature extraction (AUFE) technique, that can ultimately be deployed for geospatial analysis, modeling, and target detection. Our proposed “SAR Unsupervised and Robust Feature ExtractoR” (SURFER) solution includes in Phase I: (1) a sound and deterministic assessment of the underlying RF phenomenology and SAR processing theoretical basis for effec ...

Due to the lack of consistency in existing automated methods for assigning VNIIRS levels to motion imagery, and the overwhelming human resources required to manually assign levels, a new method of automated/semi-automated VNIIRS assessment is needed. In recent years, advancements in deep learning have provided solutions to previously intractable computer vision problems. In many cases, automated d ...

With the ever-growing number of imaging satellites in orbit the job of an analyst will change from eyes on pixels to analysis of informationfrom imagery thanks to automated image processing like object detection and change detection. Object detection algorithms have advancedto near human performance given that there is sufficient labeled data on which to train; however, obtaining this data is cost ...

This SBIR Phase II project proposes a probabilistic approach to determine a vehicle’s location using onboard video and Lidar sensors and foundation map data in GPS denied environments. The proposed system does not rely on only one type of information source, instead it combines proposals from a variety of location estimators to find a vehicles location in GPS-denied environments. The system take ...

With respect to digital imaging sensors and systems, obtaining as many pixels on target is a necessity to classify or identify a target for real-time operations and forensic analysis within the intelligence community. Rather than relying on improved sensors, the approach originally solicited and further refined here utilizes super-resolution image processing techniques to provide more detail in th ...

This SBIR Phase II project will develop biologically inspired computational models and algorithms to enable low-shot and one-shot detectionof objects-of-interest in remote sensing imagery. The Phase II effort will build upon our Phase I work including multi-scale representationlearning framework and deep-learning based feature extraction and matching techniques for low-shot target detection. The P ...