SGFormer: Satellite-Ground Fusion for 3D Semantic Scene Completion [CVPR'25]

🌐Project page | 📝Paper

Xiyue Guo¹ · Jiarui Hu¹ · Junjie Hu² · Hujun Bao¹ · Guofeng Zhang^1*
1 State Key Lab of CAD&CG, Zhejiang University, ² Chinese Universty of Hangkang, Shenzhen
^* Corresponding author.

This is the official implementation of SGFormer: Satellite-Ground Fusion for 3D Semantic Scene Completion. SGFormer is the first satellite-ground cooperative SSC framework that achieves state-of-the-art performance in scene semantic completion.

Training

Download the pretrained weight of the satellite backbone Semantic-KITTI: Google Drive

python train.py 

Eval

Download the weight of our model Semantic-KITTI: Google Drive

python eval.py 

Visualization Result

News

• 2025.02.26 --- Our paper has been accepted at CVPR 2025！
• 2025.04.08 --- We have updated the README.md and are preparing to open-source our code！
• 2025.08.25 --- We have update our code