Thesis - AI-Driven Feature Detection, Matching, and Efficient Model Deployment for Space Applications. In computer vision, the Feature Detection and Matching (FDM) task has been widely developed and innovated over the years, achieving the potential to be applied in many use cases. Its primary objective is to extract and establish accurate and reliable feature correspondences between different images, forming the foundations for various applications, including autonomous visual navigation, 3D reconstruction, object detection and tracking. This thesis is dedicated to testing and validating FDM algorithms in space-related contexts, focusing on Rendezvous and Proximity Operations (RPO). In this scenario, a chaser spacecraft acquires and processes visual information of a target spacecraft, relying on features matching for localization and navigation.