Ensemble-based Semi-supervised Learning to Improve Noisy Soiling Annotations in Autonomous Driving
This addresses the challenge of expensive and variable manual annotation for soiling categories like water drops or mud particles in autonomous driving, though it is incremental as it builds on existing pseudo-label and ensemble techniques.
The paper tackles the problem of noisy annotations for soiling segmentation in autonomous driving by using an ensemble-based semi-supervised learning method to refine labels, resulting in significant improvements in model performance.
Manual annotation of soiling on surround view cameras is a very challenging and expensive task. The unclear boundary for various soiling categories like water drops or mud particles usually results in a large variance in the annotation quality. As a result, the models trained on such poorly annotated data are far from being optimal. In this paper, we focus on handling such noisy annotations via pseudo-label driven ensemble model which allow us to quickly spot problematic annotations and in most cases also sufficiently fixing them. We train a soiling segmentation model on both noisy and refined labels and demonstrate significant improvements using the refined annotations. It also illustrates that it is possible to effectively refine lower cost coarse annotations.