Self-Supervised Uncalibrated Multi-View Video Anonymization in the Operating Room

Keqi Chen, Vinkle Srivastav, Armine Vardazaryan, Cindy Rolland, Didier Mutter, Nicolas Padoy

Introduction

Privacy preservation is a prerequisite for using video data in Operating Room (OR) research. Effective anonymization relies on the exhaustive localization of every individual; even a single missed detection necessitates extensive manual correction. However, existing approaches face two critical scalability bottlenecks: (1) they usually require manual annotations of each new clinical site for high accuracy; (2) while multi-camera setups have been widely adopted to address single-view ambiguity, camera calibration is typically required whenever cameras are repositioned. To address these problems, we propose a novel self-supervised multi-view video anonymization framework consisting of whole-body person detection and whole-body pose estimation, without annotation or camera calibration. Our core strategy is to enhance the single-view detector by "retrieving" false negatives using temporal and multi-view context, and conducting self-supervised domain adaptation. We first run an off-the-shelf whole-body person detector in each view with a low-score threshold to gather candidate detections. Then, we retrieve the low-score false negatives that exhibit consistency with the high-score detections via tracking and self-supervised uncalibrated multi-view association. These recovered detections serve as pseudo labels to iteratively fine-tune the whole-body detector. Finally, we apply whole-body pose estimation on each detected person, and fine-tune the pose model using its own high-score predictions. Experiments on the 4D-OR dataset of simulated surgeries and our dataset of real surgeries show the effectiveness of our approach achieving over 97% recall. Moreover, we train a real-time whole-body detector using our pseudo labels, achieving comparable performance and highlighting our method's practical applicability.

In this repo we provide:

• Training and inference code for video anonymization in the 4D-OR dataset.
• Trained models on the 4D-OR dataset.

Installation

1. Clone this repo, and we'll call the directory that you cloned as ${ROOT_DIR}.
2. Install dependencies.

> conda create -n anonymization python=3.10
> conda activate anonymization
(anonymization)> conda install pytorch==2.5.1 torchvision==0.20.1 pytorch-cuda=11.8 -c pytorch -c nvidia
(anonymization)> pip install -r requirements.txt

3. Install Torchreid following deep-person-reid.
4. Install Iter-Deformable-DETR following p-d-detr.
5. Install MMPOSE following mmpose.
6. Install DEIM following deim.

Data preparation

4D-OR dataset

1. Download the 4D-OR dataset and place it in ./data/ as:

${ROOT_DIR}
|-- data
    |-- 4D-OR
        |-- export_holistic_take1_processed
            |-- colorimage

2. Run command:

python ./data/generate_img_dicts.py

Trained models

Download the trained models (P-D-DETR, Multi-View Association, and DEIM respectively) as follows for direct inference.

> wget https://s3.unistra.fr/camma_public/github/OR_anonymization/4dor_iter1.pth
> wget https://s3.unistra.fr/camma_public/github/OR_anonymization/4dor_iter1_mva.pth.tar
> wget https://s3.unistra.fr/camma_public/github/OR_anonymization/deim_hgnetv2_x_4dor.pth

Training

4D-OR dataset

bash scripts/train_4dor.sh ../data/4D-OR/export_holistic_take1_processed ../data/4D-OR/detections/export_holistic_take1_processed.lmdb export_holistic_take1_processed 0

Inference

4D-OR dataset

bash scripts/inference_4dor.sh ../data/4D-OR/export_holistic_take1_processed ../data/4D-OR export_holistic_take1_processed_iter1 0

Citation

If you use our code or models in your research, please cite with:

@article{chen2026self,
  title={Self-Supervised Uncalibrated Multi-View Video Anonymization in the Operating Room},
  author={Chen, Keqi and Srivastav, Vinkle and Vardazaryan, Armine and Rolland, Cindy and Mutter, Didier and Padoy, Nicolas},
  journal={arXiv preprint arXiv:2602.02850},
  year={2026}
}

References

The project uses deep-person-reid, p-d-detr, mmpose, and deim. We thank the authors for releasing their codes.

License

This code and models are available for non-commercial scientific research purposes as defined in the CC BY-NC-SA 4.0. By downloading and using this code you agree to the terms in the LICENSE. Third-party codes are subject to their respective licenses.