# Depth Anything 3 ROS 2 Wrapper Development

**URL:** <https://discourse.openrobotics.org/t/depth-anything-3-ros-2-wrapper-development/51030>\
**Category:** ROS General\
**Created:** [November 17, 2025, 10:56pm UTC](https://discourse.openrobotics.org/t/depth-anything-3-ros-2-wrapper-development/51030 "2025-11-17T22:56:42Z")\
**Posts on this page:** 6\
**Page:** 1

<div class="post-metadata">

**Author:** ![GerdsenAI](https://sea2.discourse-cdn.com/flex022/user_avatar/discourse.openrobotics.org/gerdsenai/32/31189_2.png) [@GerdsenAI](https://discourse.openrobotics.org/u/GerdsenAI)\
**Post date:** [November 17, 2025, 10:56pm UTC](https://discourse.openrobotics.org/t/depth-anything-3-ros-2-wrapper-development/51030/1 "2025-11-17T22:56:42Z")

</div>

_edited: fix link for wrapper repo_

This is a repost from [openrobotics.zulipchat.com](http://openrobotics.zulipchat.com) that I made earlier, it was suggested I post here, so here goes…

So, over the past couple days we’ve been working on getting Depth Anything 3 (DA3 - the new monocular depth estimation model from ByteDance) running with ROS2. For those unfamiliar, Depth Anything 3 is basically a neural network that can estimate depth from a single camera image - no stereo rig or LiDAR needed. It’s pretty impressive compared to older methods like MiDaS.

**Depth Anything V3 paper:** [[2408.02532] Singularity categories of rational double points in arbitrary characteristic](https://arxiv.org/abs/2408.02532)  
**Official DA3 repo:** [GitHub - ByteDance-Seed/Depth-Anything-3: Depth Anything 3](https://github.com/ByteDance-Seed/Depth-Anything-3)  
**Our GitHub DA3 ROS2 Wrapper Repo:**

Here’s what the system looks like running on our Jetson: [GitHub - GerdsenAI/GerdsenAI-Depth-Anything-3-ROS2-Wrapper: ROS2 wrapper for Depth Anything 3 (https://github.com/ByteDance-Seed/Depth-Anything-3)](https://github.com/GerdsenAI/GerdsenAI-Depth-Anything-3-ROS2-Wrapper)

 ![DA3_wrapper_working_pipeline](https://us1.discourse-cdn.com/flex022/uploads/ros/original/3X/c/5/c5bd304e5d0c88509bb1d9513ff024c80d4cd065.jpeg)

 ![DA3_wrapper_working_pipeline2](https://us1.discourse-cdn.com/flex022/uploads/ros/original/3X/a/c/ac5af0cf0dc42dcecb8e43cbd199088c0f9d6e71.jpeg)

You can see three terminals:

- Left: USB camera node publishing at 640x480 @ 30 FPS
- Middle: Depth estimation running with the colored depth output
- Right: Depth viewer displaying the results

The depth visualization uses a turbo colormap (blue = close, red/orange = far). The quality is honestly better than we expected for monocular depth.

**Platform:** NVIDIA Jetson AGX Orin 64GB (Syslogic A4AGX64 industrial variant)

- **OS:** Ubuntu 22.04 + JetPack 6.2.1
- **CUDA:** 12.6
- **ROS2:** Humble

**Camera:** Anker PowerConf C200 2K USB webcam

- Running at 640x480 resolution
- 30 FPS output (though depth processing can’t keep up, feel free to help!! 🙂 )

**Software:**

- PyTorch 2.8.0 (Jetson-optimized from nvidia-ai-lab)
- Depth Anything 3 SMALL model (25M parameters)
- Standard v4l2\_camera for USB input

Current Performance (This is Where We Need Help)

Here’s what we’re seeing:

**Inference Performance:**

- **FPS:** 6.35 (way slower than we hoped)
- **Inference time:** 153ms per frame
- **GPU utilization:** 35-69%
- **RAM usage:** ~6 GB (out of 64 GB available)

**Is PyTorch the problem?** We’re running standard PyTorch with CUDA. Would TensorRT conversion give us a significant speedup? Has anyone done DA3 → TensorRT on Jetson?

**Memory bandwidth?** Could we be hitting memory bandwidth limits moving tensors around?

**Is the model just too big for real-time?** The SMALL model is 25M params. Maybe we need to quantize to FP16 or INT8?

**FP16 precision** - The Ampere GPU supports FP16 tensor cores. Depth estimation might not need FP32 precision.

**Optimize the preprocessing** - Right now we’re doing image normalization and resizing in Python/PyTorch. Could we push this to GPU kernels?

**Has anyone done any of this successfully?** Especially interested if anyone’s gotten DA3 or similar transformers running fast on Jetson.

The paper claims real-time performance but they’re probably testing on desktop GPUs. Getting this fast on embedded hardware is the challenge.

But, we got it working, which is cool, but 6 FPS is pretty far from real-time for most robotics applications. We’re probably doing something obviously wrong or inefficient - this is our first attempt at deploying a transformer model on Jetson.

**Contact:** [GerdsenAI · GitHub](https://github.com/GerdsenAI)  
**License:** MIT

Feel free to contribute! 🙂

---

<div class="post-metadata">

**Author:** ![GerdsenAI](https://sea2.discourse-cdn.com/flex022/user_avatar/discourse.openrobotics.org/gerdsenai/32/31189_2.png) [@GerdsenAI](https://discourse.openrobotics.org/u/GerdsenAI)\
**Post date:** [November 24, 2025, 6:17pm UTC](https://discourse.openrobotics.org/t/depth-anything-3-ros-2-wrapper-development/51030/2 "2025-11-24T18:17:15Z")

</div>

Just following up…. Has anyone had a chance to test the wrapper, or implement it into a project yet?

---

<div class="post-metadata">

**Author:** ![Phocidae](https://sea2.discourse-cdn.com/flex022/user_avatar/discourse.openrobotics.org/phocidae/32/31556_2.png) [@Phocidae](https://discourse.openrobotics.org/u/Phocidae)\
**Post date:** [December 8, 2025, 10:44am UTC](https://discourse.openrobotics.org/t/depth-anything-3-ros-2-wrapper-development/51030/3 "2025-12-08T10:44:30Z")

</div>

Hi there,

thank you very much for your contribution! Currently I am using da3 and your wrapper for my project, I try to apply it for obstacle detection and collide avoidance for my robot, I am running it on rtx 5060 and the result is very promising.

Though I am still wondering how you do the scale alignment? Could you share some experience with me about that? I am currently using piecewise scale alignment, i.e. I fit a linear function d\_gt = a\*d\_mde+b for each depth interval. What is your opinion about that?

Looking forward to your reply, thank you 🙂

---

<div class="post-metadata">

**Author:** ![AljazJus](https://sea2.discourse-cdn.com/flex022/user_avatar/discourse.openrobotics.org/aljazjus/32/31741_2.png) [@AljazJus](https://discourse.openrobotics.org/u/AljazJus)\
**Post date:** [January 4, 2026, 3:18pm UTC](https://discourse.openrobotics.org/t/depth-anything-3-ros-2-wrapper-development/51030/4 "2026-01-04T15:18:21Z")

</div>

Hi,

thank you for the repository, I have tested it out on Jetson Nano Orion (8GB RAM) and it works. But I’m trying to increase the performance and I’m having trouble with the optimisation script. Currently i get the forward pass takes around 0.12 s. I tried the scripts you have provided and had many problem some with loading the models and some with optimising. Do you have nay tips, i have ben able to quantise the model but any optimisation using TensorRT fails and couldn’t get ti to work.  
Otherwise the package was very great and useful. Thanks 🙂

---

<div class="post-metadata">

**Author:** ![GerdsenAI](https://sea2.discourse-cdn.com/flex022/user_avatar/discourse.openrobotics.org/gerdsenai/32/31189_2.png) [@GerdsenAI](https://discourse.openrobotics.org/u/GerdsenAI)\
**Post date:** [February 5, 2026, 3:28am UTC](https://discourse.openrobotics.org/t/depth-anything-3-ros-2-wrapper-development/51030/5 "2026-02-05T03:28:24Z")

</div>

Sorry, just getting back to this @AljazJus . I have TensorRT working, it’s MUCH quicker. Will be making a post shortly.

---

<div class="post-metadata">

**Author:** ![GerdsenAI](https://sea2.discourse-cdn.com/flex022/user_avatar/discourse.openrobotics.org/gerdsenai/32/31189_2.png) [@GerdsenAI](https://discourse.openrobotics.org/u/GerdsenAI)\
**Post date:** [February 5, 2026, 4:58pm UTC](https://discourse.openrobotics.org/t/depth-anything-3-ros-2-wrapper-development/51030/6 "2026-02-05T16:58:11Z")

</div>

Great news for everyone following this project! We’ve successfully implemented TensorRT 10.3 acceleration, and the results are significant:

### Performance Improvement

| **Metric** | **Before (PyTorch)** | **After (TensorRT)** | **Improvement** |
| --- | --- | --- | --- |
| FPS | 6.35 | **43+** | **6.8x faster** |
| Inference Time | 153ms | ~23ms | **6.6x faster** |
| GPU Utilization | 35-69% | 85%+ | More efficient |

**Test Platform:** Jetson Orin NX 16GB ([Seeed reComputer J4012](https://www.seeedstudio.com/reComputer-Robotics-J4012-with-GMSL-extension-board-p-6537.html)), JetPack 6.2, TensorRT 10.3

### Key Technical Achievement: Host-Container Split Architecture

We solved a significant Jetson deployment challenge - TensorRT Python bindings are broken in current Jetson container images ([dusty-nv/jetson-containers#714](https://github.com/dusty-nv/jetson-containers/issues/714)). Our solution:

```auto
HOST (JetPack 6.x)
+--------------------------------------------------+
| TRT Inference Service (trt_inference_shm.py) |
| - TensorRT 10.3, ~15ms inference |
+--------------------------------------------------+
                    ↑
                    | /dev/shm/da3 (shared memory, ~8ms IPC)
                    ↓
+--------------------------------------------------+
| Docker Container (ROS2 Humble) |
| - Camera drivers, depth publisher |
+--------------------------------------------------+

```

This architecture enables real-time TensorRT inference while keeping ROS2 in a clean container environment.

### Call for Contributors

We’re looking for help with:

- **Test coverage** for SharedMemory/TensorRT code paths

- **Validation on other Jetson platforms** (AGX Orin, Orin Nano)

- **Point cloud generation** (currently depth-only)

**Repo:** [https://github.com/GerdsenAI/GerdsenAI-Depth-Anything-3-ROS2-Wrapper](https://github.com/GerdsenAI/GerdsenAI-Depth-Anything-3-ROS2-Wrapper)  
**License:** MIT

@Phocidae @AljazJus - the TensorRT optimization should help significantly with your projects! Let me know if you run into any issues.