Deep Learning Training Accelerator (DeLTA) for Space Object Detection

Deep Learning Training Accelerator (DeLTA) for Space Object Detection is an inline approach to reduce the amount of data required to train object recognition models without significant loss of performance; thus, cutting respective training time. DeLTA addresses the industry recognized challenge of costly  - in terms of time and computing resources - Deep Neural Network (DNN)  model training performed to achieve optimal precision and recall. The near term impact of a successful effort would drastically improve object recognition system development timelines  and processes in the space arena.  Moreover, because it considers generality across multiple sensors, convolution neural network (CNN) models, and data sets as a fundamental conceptual requirement, DeLTA can be applied to various other markets.