Skip to content

yty-sky/AdaptiveOcc

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

AdaptiveOcc

AdaptiveOcc: Adaptive Octree-based Network for Multi-Camera 3D Semantic Occupancy Prediction

News

  • 2024-11-05: We release the code.
  • 2024-11-03: Our paper is accepted by TCSVT.

Introduction

In this paper, we propose AdaptiveOcc, a novel octree-based multi-level architecture designed for multi-camera 3D semantic occupancy prediction. It can adaptively represent different parts of space with varying voxel granularity. The adaptability of AdaptiveOcc is highlighted by its capability to intelligently output large homogeneous voxel blocks at shallower layers and ensure the propagation of more intricate voxel structures at deeper layers, enhancing fine-grained perception of critical regions. Futhermore, We propose a distance-adaptive octree construction rule for generating supervised labels to endow our model with adaptability. Considering that the voxel granularity requirements vary for different distance ranges in environmental perception, such a construction rule results in a higher likelihood of coarser granularity for distant regions and finer granularity for nearby regions. This ensures a more efficient and rational allocation of computational resources, further reducing the inference latency.

Highlights

Multi-granularity scene representation

Relit

We propose a distance-adaptive octree construction rule to representation scenes using multi-granularity voxels.

Multi-level hierarchical model

Pipeline

We propose a novel octree-based multi-level hierarchical model, which can adaptively represent different parts of the space with varying voxel granularity.

Strong resolution scalability

main_res main_res

Our method exhib superior performance in terms of resolution scalability, which can scale to finer granularities with faster speed, and less training memory compared with other state-of-the-art methods.

Getting Started

Acknowledgement

Many thanks to these excellent open-source projects:

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published