Machine learning based object detection using radar information

An apparatus for detecting objects, comprising:

• a memory;
• and
• one or more processors coupled to the memory and configured to:
• obtain radar information identifying measured properties of at least one object in an environment;
• generate pre-processed radar information for input into a neural network;
• generate an object detection output for the at least one object at least in part by detecting the at least one object using the neural network with the pre-processed radar information as input, wherein the object detection output of the neural network comprises a bounding region identifying a location of the at least one object;
• determine that a size of the bounding region is less than a threshold size;
• and
• modifying, based on the radar information and the size of the bounding region being less than the threshold size, a classification confidence of the object detection output for the at least one object.