Method and system for training machine learning algorithm to detect objects at distance

This invention describes a method to teach a machine learning system to detect objects at a greater distance than it was initially trained for. It works by aligning close-up sensor data, where objects are already recognized, with new, further-away sensor data. The system then uses the recognized objects from the close-up data to label corresponding parts of the far-away data, and retrains itself to identify objects at the increased distance, specifically using timestamps and locations for alignment and object classes from a 3D map.