Teachers in concordance for pseudo-labeling of 3D sequential data
This addresses the challenge of reducing manual labeling effort in safety-critical autonomous driving applications, though it is incremental as it builds on existing pseudo-labeling techniques.
The paper tackles the problem of leveraging unlabeled sequential 3D data for autonomous driving by proposing a teacher-student pseudo-labeling method using multiple teachers with different temporal information, which outperforms some fully supervised methods using only 20% manual labels and improves performance for rare classes.
Automatic pseudo-labeling is a powerful tool to tap into large amounts of sequential unlabeled data. It is specially appealing in safety-critical applications of autonomous driving, where performance requirements are extreme, datasets are large, and manual labeling is very challenging. We propose to leverage sequences of point clouds to boost the pseudolabeling technique in a teacher-student setup via training multiple teachers, each with access to different temporal information. This set of teachers, dubbed Concordance, provides higher quality pseudo-labels for student training than standard methods. The output of multiple teachers is combined via a novel pseudo label confidence-guided criterion. Our experimental evaluation focuses on the 3D point cloud domain and urban driving scenarios. We show the performance of our method applied to 3D semantic segmentation and 3D object detection on three benchmark datasets. Our approach, which uses only 20% manual labels, outperforms some fully supervised methods. A notable performance boost is achieved for classes rarely appearing in training data.