CVSformer: Cross-View Synthesis Transformer for Semantic Scene Completion

TL;DR

CVSformer introduces a novel transformer-based approach for semantic scene completion that effectively models cross-view object relationships, leading to state-of-the-art results in 3D scene understanding.

Contribution

The paper proposes CVSformer, a new model combining multi-view feature synthesis and cross-view transformer to improve 3D scene completion accuracy.

Findings

Achieves state-of-the-art performance on public datasets.

Effectively models cross-view relationships for occluded object reasoning.

Outperforms existing voxel-based SSC methods.

Abstract

Semantic scene completion (SSC) requires an accurate understanding of the geometric and semantic relationships between the objects in the 3D scene for reasoning the occluded objects. The popular SSC methods voxelize the 3D objects, allowing the deep 3D convolutional network (3D CNN) to learn the object relationships from the complex scenes. However, the current networks lack the controllable kernels to model the object relationship across multiple views, where appropriate views provide the relevant information for suggesting the existence of the occluded objects. In this paper, we propose Cross-View Synthesis Transformer (CVSformer), which consists of Multi-View Feature Synthesis and Cross-View Transformer for learning cross-view object relationships. In the multi-view feature synthesis, we use a set of 3D convolutional kernels rotated differently to compute the multi-view features for each voxel. In the cross-view transformer, we employ the cross-view fusion to comprehensively learn the cross-view relationships, which form useful information for enhancing the features of individual views. We use the enhanced features to predict the geometric occupancies and semantic labels of all voxels. We evaluate CVSformer on public datasets, where CVSformer yields state-of-the-art results.