ETA: Energy-based Test-time Adaptation for Depth Completion

  • Chung, Younjoon; 
  • Park, Hyoungseob; 
  • Rim, Patrick; 
  • Zhang, Xiaoran; 
  • He, Jihe; 
  • ... Hong, Byung-Woo; 
  • 외 4명
Citations

WEB OF SCIENCE

2

초록

We propose a method for test-time adaptation of pretrained depth completion models. Depth completion models, trained on some "source" data, often predict erroneous outputs when transferred to "target" data captured in novel environmental conditions due to a covariate shift. The crux of our method lies in quantifying the likelihood of depth predictions belonging to the source data distribution. The challenge is in the lack of access to out-of-distribution (target) data prior to deployment. Hence, rather than making assumptions regarding the target distribution, we utilize adversarial perturbations as a mechanism to explore the data space. This enables us to train an energy model that scores local regions of depth predictions as in- or out-of-distribution. We update the parameters of pretrained depth completion models at test time to minimize energy, effectively aligning test-time predictions to those of the source distribution. We call our method "Energy-based Test-time Adaptation", or ETA for short. We evaluate our method across three indoor and three outdoor datasets, where ETA improve over the previous state-of-the-art method by an average of 6.94% for outdoors and 10.23% for indoors.

키워드

PROPAGATION NETWORK; SPARSE
제목
ETA: Energy-based Test-time Adaptation for Depth Completion
저자
Chung, Younjoon; Park, Hyoungseob; Rim, Patrick; Zhang, Xiaoran; He, Jihe; Zeng, Ziyao; Cicek, Safa; Hong, Byung-Woo; Duncan, James S.; Wong, Alex
DOI
10.1109/ICCV51701.2025.00567
발행일
2025
유형
Proceedings Paper
저널명
Proceedings of the IEEE International Conference on Computer Vision
페이지
6001 ~ 6012