Award Data

The goal of this project is to develop an attention-guided pattern and geospatial analysis system for imagery exploitation. This new technology will be based on the dynamic receptive field model of visual attention that has emerged from neuroscienceresearch. The attention system holds the key to the success of the vision systems of many higher form animals struggling to survive: They see what they ...

During Phase II, Spectral Sciences Inc., proposes to develop innovative software tools for timely and highly specific characterization of the world littoral zones using data from new satellite-based hyperspectral imaging systems. SS Phase II objectivesinclude (1) develop a software package containing an automated LZ atmospheric correction code with advanced algorithms for determining ocean bot ...

Next Century Corporation proposes the development of Muggsy, a low-shot deep learning detection prototype system that learns to recognize uncommon targets in remote imagery. Our Phase I research extends and leverages an image classification system of our own design called EvoDevo. EvoDevo evolves its own neural network architecture before training to meet the complexity of the data. Muggsy uses le ...

To detect uncommon targets in remote sensing imagery, it is quite often that very few prior examples are available. This so-called low-shot detection remains a very challenging problem in remote sensing, despite the recent development in state-of-the-art object detection algorithms such as Faster R-CNN and YOLO, and low-shot learning methods such as feature shrinking, model regression and memory a ...

Ground vehicles with navigation capability (e.g., GPS) can index into foundation data (e.g., Google Maps) to gain situational awareness abouttheir surroundings. When GPS and RF navigation sources are degraded, maintaining situational awareness requires an alternative navigationsource. One alternative source is the foundation data itself. The data contain objects at known 3D locations, which projec ...

Topographic features found in ground-based natural images contain information that is useful for a variety of applications including locationestimation and navigation. Traditionally these features have been manually labeled by analysts which is costly and time consuming, especiallyconsidering the volume of readily available data. We propose a novel method for extracting topographic features from s ...

Next Century Corporation proposes to create TopoRobo (topographical annotation robot software) to automatically extract depth-orderedlists of ridge polylines overlaid on an image or video mosaics to feed topography-based geolocation algorithms. TopoRobo will leverage deepneural network machine learning methods optimized for topographical features through EvoDevo, Next Centurys algorithm to grow an ...

With the explosive growth of Internet of Things (IoT) and mobile communication technologies, media streaming service and applications basedon video content have gained remarkable popularity and interest from users. When someone is using their device to record a video, then sharethat video with friends through a certain website (e.g. Netflix, YouTube), the process may sound simple from the user sid ...

Following technology advances in high-performance computation systems and fast growth of data acquisition, a technical breakthroughnamed Deep Learning made remarkable success in many research areas and applications. Nevertheless, the progress of hardwaredevelopment still falls far behind the upscaling of deep neural network (DNN) models at the software level. NGA seeks to apply neuralnetwork minia ...

Generalized change detection is a critical capability to mitigate the need for massive human inspection of the rapidly expanding volume ofglobal overhead satellite imagery. Current optical change detection approaches focus on fully specified systems to detect a predefined set ofchanges, and effective approaches for generalized change detection have not yet been demonstrated. We propose to build a ...