Monitored Distillation for Positive Congruent Depth Completion
- Authors
- Liu, T.Y.; Agrawal, P.; Chen, A.; Hong, Byung-Woo; Wong, A.
- Issue Date
- Oct-2022
- Publisher
- Springer Science and Business Media Deutschland GmbH
- Keywords
- Blind ensemble; Depth completion; Knowledge distillation
- Citation
- Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), v.13662 LNCS, pp 35 - 53
- Pages
- 19
- Journal Title
- Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
- Volume
- 13662 LNCS
- Start Page
- 35
- End Page
- 53
- URI
- https://scholarworks.bwise.kr/cau/handle/2019.sw.cau/59679
- DOI
- 10.1007/978-3-031-20086-1_3
- ISSN
- 0302-9743
1611-3349
- Abstract
- We propose a method to infer a dense depth map from a single image, its calibration, and the associated sparse point cloud. In order to leverage existing models (teachers) that produce putative depth maps, we propose an adaptive knowledge distillation approach that yields a positive congruent training process, wherein a student model avoids learning the error modes of the teachers. In the absence of ground truth for model selection and training, our method, termed Monitored Distillation, allows a student to exploit a blind ensemble of teachers by selectively learning from predictions that best minimize the reconstruction error for a given image. Monitored Distillation yields a distilled depth map and a confidence map, or “monitor”, for how well a prediction from a particular teacher fits the observed image. The monitor adaptively weights the distilled depth where if all of the teachers exhibit high residuals, the standard unsupervised image reconstruction loss takes over as the supervisory signal. On indoor scenes (VOID), we outperform blind ensembling baselines by 17.53% and unsupervised methods by 24.25%; we boast a 79% model size reduction while maintaining comparable performance to the best supervised method. For outdoors (KITTI), we tie for 5th overall on the benchmark despite not using ground truth. Code available at: https://github.com/alexklwong/mondi-python. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.
- Files in This Item
- There are no files associated with this item.
- Appears in
Collections - College of Software > Department of Artificial Intelligence > 1. Journal Articles
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.