# DART: Depth-as-Target Pretraining for Surgical Vision Foundation Models

arXiv: https://arxiv.org/abs/2609.04555 · HTML: https://mlsanity.com/paper/2609.04555/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 45 of its 53 references that are in the map.
- First posted on arXiv 3 Sep 2026.

## Builds on (most cited first, 30 of 45)

- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [Masked Autoencoders Are Scalable Vision Learners](https://mlsanity.com/paper/2111.06377/) (2021)
- [Vision Transformers for Dense Prediction](https://mlsanity.com/paper/2103.13413/) (2021)
- [Depth Anything V2](https://mlsanity.com/paper/2406.09414/) (2024)
- [DINOv3](https://mlsanity.com/paper/2508.10104/) (2025)
- [Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecture](https://mlsanity.com/paper/2301.08243/) (2023)
- [iBOT: Image BERT Pre-Training with Online Tokenizer](https://mlsanity.com/paper/2111.07832/) (2021)
- [EndoNet: A Deep Architecture for Recognition Tasks on Laparoscopic Videos](https://mlsanity.com/paper/1602.03012/) (2016)
- [Masked Siamese Networks for Label-Efficient Learning](https://mlsanity.com/paper/2204.07141/) (2022)
- [MultiMAE: Multi-modal Multi-task Masked Autoencoders](https://mlsanity.com/paper/2204.01678/) (2022)
- [Neural Rendering for Stereo 3D Reconstruction of Deformable Tissues in Robotic Surgery](https://mlsanity.com/paper/2206.15255/) (2022)
- [CholecSeg8k: A Semantic Segmentation Dataset for Laparoscopic Cholecystectomy Based on Cholec80](https://mlsanity.com/paper/2012.12453/) (2020)
- [Endo-Depth-and-Motion: Reconstruction and Tracking in Endoscopic Videos using Depth Networks and Photometric Constraints](https://mlsanity.com/paper/2103.16525/) (2021)
- [Foundation Model for Endoscopy Video Analysis via Large-scale Self-supervised Pre-train](https://mlsanity.com/paper/2306.16741/) (2023)
- [Learning Multi-modal Representations by Watching Hundreds of Surgical Video Lectures](https://mlsanity.com/paper/2307.15220/) (2023)
- [Pri3D: Can 3D Priors Help 2D Representation Learning?](https://mlsanity.com/paper/2104.11225/) (2021)
- [SAR-RARP50: Segmentation of surgical instrumentation and Action Recognition on Robot-Assisted Radical Prostatectomy Challenge](https://mlsanity.com/paper/2401.00496/) (2023)
- [Surgical-DINO: Adapter Learning of Foundation Models for Depth Estimation in Endoscopic Surgery](https://mlsanity.com/paper/2401.06013/) (2024)
- [Towards Unified Surgical Skill Assessment](https://mlsanity.com/paper/2106.01035/) (2021)
- [Dissecting Self-Supervised Learning Methods for Surgical Computer Vision](https://mlsanity.com/paper/2207.00449/) (2022)
- [HecVL: Hierarchical Video-Language Pretraining for Zero-shot Surgical Phase Recognition](https://mlsanity.com/paper/2405.10075/) (2024)
- [Surgical Robot Transformer (SRT): Imitation Learning for Surgical Tasks](https://mlsanity.com/paper/2407.12998/) (2024)
- [Tracking and Mapping in Medical Computer Vision: A Review](https://mlsanity.com/paper/2310.11475/) (2023)
- [SRT-H: A Hierarchical Framework for Autonomous Surgery via Language Conditioned Imitation Learning](https://mlsanity.com/paper/2505.10251/) (2025)
- [General surgery vision transformer: A video pre-trained foundation model for general surgery](https://mlsanity.com/paper/2403.05949/) (2024)
- [Scaling up self-supervised learning for improved surgical foundation models](https://mlsanity.com/paper/2501.09436/) (2025)
- [Data Splits and Metrics for Method Benchmarking on Surgical Action Triplet Datasets](https://mlsanity.com/paper/2204.05235/) (2022)
- [Self-Supervised Learning for Endoscopic Video Analysis](https://mlsanity.com/paper/2308.12394/) (2023)
- [Dynamic Scene Graph Representation for Surgical Video](https://mlsanity.com/paper/2309.14538/) (2023)

## Abstract (arXiv)

Vision foundation models (VFMs) are valuable in data-scarce domains such as surgery, where a single pretrained backbone can provide rich representations for many downstream tasks. Yet the dominant self-supervised pretraining paradigm uses only RGB images, leaving readily available complementary signals, such as depth maps, unused. This is a particular missed opportunity in surgery, where natural-image VFMs transfer poorly while the scene geometry is rich and informative. With strong off-the-shelf models now able to produce pseudo-labeled dense depth for any image corpus, we hypothesize that such signals can be folded into pretraining to learn better representations. We present DART, an RGB-D pretraining recipe that builds on DINOv2 with a simple modification: a pixel-space depth reconstruction objective applied to masked iBOT patches, supervised by pseudo-labeled depth. Depth is used only during pretraining, so fine-tuning and inference remain RGB-only. We find that this pixel-level reconstruction head improves representation quality rather than disrupting it. We further show that depth, which encodes scene geometry, is more effective as a target than alternative dense signals such as Canny edges, confirming that the gains stem from depth rather than added supervision alone. Across eight surgical benchmarks spanning segmentation, depth estimation, and image-level recognition, DART outperforms both natural-image and in-domain baselines, including a vanilla DINOv2 trained on identical data, improving dense prediction while also strengthening image-level understanding. More broadly, DART shows that freely available geometric pseudo-labels can strengthen foundation model pretraining without extra labels or added inference cost, pointing toward stronger backbones for surgery.