Low-Shot Detection in Remote Sensing Imagery

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 Phase II effort willfocus towards improvement of the Phase I technologies and transitioning them into a prototype system including an end-to-end software withtraining and testing capabilities of target detection in remote sensing data. The Phase II effort will leverage Novateur Teams expertise in theareas of deep learning and convolutional neural network, metric and hybrid learning, domain adaptation, top-down and bottom-up visualattention, exploitation of remote sensing imagery, and image and scene understanding.