Semantic Label Reduction Techniques for Autonomous Driving
This work addresses efficiency in autonomous driving systems by reducing unnecessary label complexity, but it appears incremental as it focuses on evaluating and remapping existing labels without introducing new methods.
The paper tackles the problem of identifying which semantic segmentation labels are necessary for autonomous driving control decisions, and finds that remapping non-critical labels to other classes simplifies the task by reducing to only critical labels.
Semantic segmentation maps can be used as input to models for maneuvering the controls of a car. However, not all labels may be necessary for making the control decision. One would expect that certain labels such as road lanes or sidewalks would be more critical in comparison with labels for vegetation or buildings which may not have a direct influence on the car's driving decision. In this appendix, we evaluate and quantify how sensitive and important the different semantic labels are for controlling the car. Labels that do not influence the driving decision are remapped to other classes, thereby simplifying the task by reducing to only labels critical for driving of the vehicle.