zh/guides/nvidia-jetson/

[giscus[bot]]

zh/guides/nvidia-jetson/

通过我们的详细指南，了解如何在NVIDIA Jetson 设备上部署Ultralytics YOLO11 。探索性能基准，最大限度地发挥人工智能功能。

https://docs.ultralytics.com/zh/guides/nvidia-jetson/

[Vectorvin]

jetson orin nano目前的部署是存在问题的。
我的JetPack版本是6.2，其中：TensorRT 10.3，cuda 12.6。使用的python版本3.10.2.
在使用官方转换格式样例时：

from ultralytics import YOLO

# Load a YOLO11n PyTorch model
model = YOLO("./model/weights/best.pt")

# Export the model to TensorRT
model.export(format="engine")  # creates 'best.engine'

会出现如下错误

requirements: Ultralytics requirement ['tensorrt>7.0.0,!=10.1.0'] not found, attempting AutoUpdate...
  error: subprocess-exited-with-error
  
  × python setup.py egg_info did not run successfully.
  │ exit code: 1
  ╰─> [6 lines of output]
      Traceback (most recent call last):
        File "<string>", line 2, in <module>
        File "<pip-setuptools-caller>", line 34, in <module>
        File "/tmp/pip-install-zgxloadv/tensorrt-cu12_1ec32cbb8cb447798ddc82b6bffd2bb5/setup.py", line 71, in <module>
          raise RuntimeError("TensorRT does not currently build wheels for Tegra systems")
      RuntimeError: TensorRT does not currently build wheels for Tegra systems
      [end of output]
 note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed

× Encountered error while generating package metadata.
╰─> See above for output.

note: This is an issue with the package mentioned above, not pip.
hint: See above for details.
Retry 1/2 failed: Command 'pip install --no-cache-dir "tensorrt>7.0.0,!=10.1.0" ' returned non-zero exit status 1.
  error: subprocess-exited-with-error

是的。这看起来是tensorRT版本不兼容的问题，但我查过JetPack6.1的版本，也是使用的TensorRT 10.3。问题应该出现在哪呢？

[glenn-jocher]

@Vectorvin 感谢您的反馈！错误表明当前安装的 TensorRT 包不支持 Jetson 系统。建议您手动安装适用于 JetPack 6.2 的 TensorRT 包，而不是通过 pip 安装。可以参考 NVIDIA Jetson TensorRT 文档 或 Ultralytics NVIDIA Jetson 指南 获取适配 Jetson 的安装方法。

[Vectorvin]

TensorRT是配套JetPack的，没法手动修改进行适配。我很不理解你们是怎么在JetPack6.1的情况下怎么做到TensorRT还能适配的

[glenn-jocher]

@Vectorvin 感谢您的反馈！在 JetPack 6.1 的环境下，我们建议严格按照 NVIDIA 官方文档提供的 TensorRT 和 PyTorch 安装指南来确保兼容性。由于 TensorRT 是随 JetPack 提供的，因此无法通过 pip 或其他方式更新到不兼容的版本。如需进一步参考，请查看 Ultralytics NVIDIA Jetson 指南。

[UltralyticsAssistant]

👋 Hello, thank you for sharing and for your interest in Ultralytics 🚀!

We noticed you're referencing NVIDIA Jetson-specific deployment 🌍. For the best insights, feel free to check out our detailed NVIDIA Jetson Guide where you’ll find the latest documentation on deploying Ultralytics YOLO models on NVIDIA Jetson devices, enabling you to maximize AI performance effectively.

If you're encountering any issues or have additional questions, here are some tips to help us assist you better:

• If this is a 🐛 Bug Report, kindly provide a minimum reproducible example to help us diagnose the issue.
• If this is about custom deployment or specific ❓ Questions, please provide detailed context, including your setup, any modifications made, and specific errors (if applicable).

Make sure your environment matches our recommended specifications for YOLO performance:

Upgrade

Keep Ultralytics YOLO up-to-date! Upgrade to the latest version:

pip install -U ultralytics

Environments

Verified environments with all dependencies pre-installed include:

• Notebooks:
• Jetson: Follow detailed instructions in the Jetson Guide.
• Other environments remain compatible, such as Google Cloud or AWS Deep Learning AMIs.

For broader reach and troubleshooting resources, we encourage connecting with the Ultralytics community:

• Join real-time discussions on Discord 🎧.
• Explore detailed threads on Discourse or contribute to our Subreddit.

Finally, an Ultralytics engineer will review your post and assist you further soon. Stay tuned! 😊