[CVPR 2026] SeeGroup: Multi-Layer Depth Estimation of Transparent Surfaces via Self-Determined Grouping

In this work, we propose SeeGroup, a multi-layer depth estimation method that allows the model itself to adaptively assign surfaces to depth maps. We formulate per-pixel multi-layer depth as a point process, treating depth layers as unordered events along each camera ray. This induces a permutation-invariant likelihood over the observed depth layers, yielding a loss that naturally supports arbitrary layer groupings. Experiments demonstrate that our method significantly advances the state of the art of multi-layer depth estimation, improving quadruplet relative depth accuracy on LayeredDepth benchmark from 61.34% to 70.09%.

If you find SeeGroup useful for your work, please consider citing our academic paper:

SeeGroup: Multi-Layer Depth Estimation of Transparent Surfaces
via Self-Determined Grouping

Hongyu Wen, Jia Deng

@misc{wen2026seegroupmultilayerdepthestimation,
      title={SeeGroup: Multi-Layer Depth Estimation of Transparent Surfaces via Self-Determined Grouping}, 
      author={Hongyu Wen and Jia Deng},
      year={2026},
      eprint={2605.28735},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2605.28735}, 
}

Setup

Install dependencies:

pip install -r requirements.txt

Download the released SeeGroup checkpoint for validation and test prediction:

bash scripts/download_seegroup_checkpoint.sh

Validation

Evaluate SeeGroup on LayeredDepth validation split with the released checkpoint:

python val.py --checkpoint-path checkpoints/seegroup.pth

Test

Run SeeGroup on LayeredDepth validation split with the released checkpoint and save the predictions:

python test.py \
  --checkpoint-path checkpoints/seegroup.pth \
  --output-dir predictions/layereddepth_test \
  --format npy

Training

Before training, download the DAV2 backbone once:

bash scripts/download_dav2_checkpoint.sh

Run single-GPU training:

python train.py

Run multi-GPU training on one machine:

torchrun --nproc_per_node=$gpus train.py

Acknowledgement

This project relies on code from Depth-Anything-V2. We thank the original authors for their excellent work.