Unsupervised Continual Semantic Adaptation through Neural Rendering

Abstract

An increasing amount of applications rely on data-driven models that are deployed for perception tasks across a sequence of scenes. Due to the mismatch between training and deployment data, adapting the model on the new scenes is often crucial to obtain good performance. In this work, we study continual multi-scene adaptation for the task of semantic segmentation, assuming that no ground-truth labels are available during deployment and that performance on the previous scenes should be maintained. We propose training a Semantic-NeRF network for each scene by fusing the predictions of a segmentation model and then using the view-consistent rendered semantic labels as pseudo-labels to adapt the model. Through joint training with the segmentation model, the Semantic-NeRF model effectively enables 2D-3D knowledge transfer. Furthermore, due to its compact size, it can be stored in a long-term memory and subsequently used to render data from arbitrary viewpoints to reduce forgetting. We evaluate our approach on ScanNet, where we outperform both a voxel-based baseline and a state-of-the-art unsupervised domain adaptation method.

Accepted by the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2023. Zhizheng Liu and Francesco Milano share first authorship. Hermann Blum and Cesar Cadena share senior authorship. 18 pages, 8 figures, 9 tables

References in corpus (6)

• Rethinking Atrous Convolution for Semantic Image Segmentation
• Instant Neural Graphics Primitives with a Multiresolution Hash Encoding
• Decomposing NeRF for Editing via Feature Field Distillation
• Panoptic Multi-TSDFs: a Flexible Representation for Online Multi-resolution Volumetric Mapping and Long-term Dynamic Scene Consistency
• ILabel: Interactive Neural Scene Labelling
• Continual Adaptation of Semantic Segmentation using Complementary 2D-3D Data Representations