Enhancing Indoor Occupancy Prediction via Sparse Query-Based Multi-Level Consistent Knowledge Distillation [paper]

RA-L 2025

• Initial commit
• Model zoo
• arXiv version

Occupancy prediction provides critical geometric and semantic understanding for robotics but faces efficiency-accuracy trade-offs. Current dense methods suffer computational waste on empty voxels, while sparse query-based approaches lack robustness in diverse and complex indoor scenes. In this paper, we propose DiScene, a novel sparse query-based framework that leverages multi-level distillation to achieve efficient and robust occupancy prediction. In particular, our method incorporates two key innovations: (1) a Multi-level Consistent Knowledge Distillation strategy, which transfers hierarchical representations from large teacher models to lightweight students through coordinated alignment across four levels, including encoder-level feature alignment, query-level feature matching, prior-level spatial guidance, and anchor-level high-confidence knowledge transfer and (2) a Teacher-Guided Initialization policy, employing optimized parameter warm-up to accelerate model convergence. Validated on the Occ-Scannet benchmark, DiScene achieves 23.2 FPS without depth priors while outperforming our baseline method, OPUS, by 36.1% and even better than the depth-enhanced version, OPUS†. With depth integration, DiScene† attains new SOTA performance, surpassing EmbodiedOcc by 3.7% with 1.62× faster inference speed. Furthermore, experiments on the Occ3D-nuScenes benchmark and in-the-wild scenarios demonstrate the versatility of our approach in various environments.

Follow instructions HERE to prepare the environment.

Please download posed_images and gathered_data from the Occ-ScanNet Benchmark and move them to data/occscannet, zip files need extraction.

Folder structure

DiScene
├── ...
├── data/
│   ├── occscannet/
│   │   ├── gathered_data/
│   │   ├── posed_images/
│   │   ├── train.txt
│   │   ├── test.txt
├── ...

1. Train different models using 8 GPUs on Occ-ScanNet Benchmark:

   # train student model
   bash dist_train.sh 8 configs/occscannet/r50/discene_960x16_student_r50.py
   
   # train teacher model
   bash dist_train.sh 8 configs/occscannet/internxl/discene_960x16_teacher_internxl.py
   
   # train distilled model (DiScene†)
   bash dist_train.sh 8 configs/occscannet/r50/discene_960x16_guided_distill_r50.py  # Please modify 'teacher_weight' in the configuration file accordingly.
   
   # training without pre-trained depth model
   # student model
   bash dist_train.sh 8 configs/occscannet/r50/discene_960x16_vanilla_r50.py
   # teacher model
   bash dist_train.sh 8 configs/occscannet/internxl/discene_960x16_teacher_vanilla_internxl.py
   # distilled model (DiScene)
   bash dist_train.sh 8 configs/occscannet/r50/discene_960x16_guided_distill_vanilla_r50.py  # Please modify 'teacher_weight' in the configuration file accordingly.

2. Evaluate model using 8 GPUs on Occ-ScanNet Benchmark:

   # evaluate distilled model (DiScene†)
   bash dist_val.sh 8 configs/occscannet/r50/discene_960x16_guided_distill_r50.py /path/to/checkpoints

Our code is developed on top of OPUS. We sincerely appreciate their amazing works.

Also, we would like to thank these excellent open source projects:

• ISO
• EmbodiedOcc
• Metric3D
• Depth Anything V2

If you find this work useful, please consider citing:

@article{li2025enhancing,
  title={Enhancing Indoor Occupancy Prediction Via Sparse Query-Based Multi-Level Consistent Knowledge Distillation},
  author={Li, Xiang and Zheng, Yupeng and Li, Pengfei and Chen, Yilun and Zhang, Ya-Qin and Ding, Wenchao},
  journal={IEEE Robotics and Automation Letters},
  year={2025},
  volume={10},
  number={11},
  pages={11690-11697},
  doi={10.1109/LRA.2025.3615532}
}