This includes low-latency detection and tracking, motion-aware vision, sensor fusion with inertial data, 3D direction-vector estimation, GPS-denied navigation, and precision landing support. * Use camera intrinsics and extrinsics to transform image-space detections into 3D direction vectors or other navigation-relevant outputs. * Benchmark and evaluate different model and algorithm families, including approaches such as CenterNet, SuperPoint, SuperGlue, optical flow, feature tracking, object detection, and lightweight embedded models. * Strong understanding of computer vision fundamentals, camera geometry, feature detection, object detection, tracking, calibration, and image-space to 3D transformations.
mehr