Space Object Recognition using Deep Learning Methods for Space Situational Awareness
Inventor
A rapid growth of space debris and objects measured with an estimated over 750,000 debris found within the orbit, and more than 20,000 Near-Earth objects (NEO). Due to the increasing prevalence of space environment use, possible collisions or approach of different objects should always be monitored to ensure the safety of the people and secure essential space satellites and shuttles. Space Situational Awareness (SSA) system requires recognition of space objects that are varied in sizes, shapes, and types. The space images are challenging because of several factors such as illumination and noise and thus make the recognition task complex. In this project, 4 solutions are proposed for space object recognition for SSA purposes:1) RGB-D Based Multimodal Convolutional Neural Networks (CNN)2) Combination of RGB based vision transformer and Depth-based End-to-End CNN3) Localization and classification of space objects using EfficientDet detector4) Stacking of CoAtNets Using Fusion of RGB and Depth ImagesIt was found that combination of fusion and stacking was able to improve classification accuracy largely compared to the baseline methods by producing an average accuracy of 89 % and average F1 score of 89 %.
