zh/hub/inference-api/

[giscus[bot]]

zh/hub/inference-api/

了解如何使用Ultralytics HUB Inference API 运行推理。包括Python 和 cURL 示例，以便快速集成。

https://docs.ultralytics.com/zh/hub/inference-api/

[sjocnjfjd]

I have trained my model,but I can not deploy it in my app,why？100 epochs completed in 0.146 hours.
Optimizer stripped from runs/detect/train/weights/last.pt, 5.5MB
Optimizer stripped from runs/detect/train/weights/best.pt, 5.5MB

Validating runs/detect/train/weights/best.pt...
Ultralytics 8.3.44 🚀 Python-3.10.12 torch-2.5.1+cu121 CUDA:0 (Tesla T4, 15102MiB)
YOLO11n summary (fused): 238 layers, 2,582,347 parameters, 0 gradients, 6.3 GFLOPs

FileNotFoundError Traceback (most recent call last)
in <cell line: 4>()
2
3 model = YOLO('https://hub.ultralytics.com/models/kYDnuTY9JD3hHyFdnc5T')
----> 4 results = model.train(data='"D://python//新建文件夹 (2)//hakusenninnshiki.v2i.yolov11//lane.yaml"', project='runs/train', name='exp', epochs=100)

6 frames
/usr/local/lib/python3.10/dist-packages/ultralytics/data/utils.py in check_det_dataset(dataset, autodownload)
327 else:
328 m += f"\nNote dataset download directory is '{DATASETS_DIR}'. You can update this in '{SETTINGS_FILE}'"
--> 329 raise FileNotFoundError(m)
330 t = time.time()
331 r = None # success

FileNotFoundError:
Dataset '/content/datasets/lane/lane.yaml' images not found ⚠️, missing path '/content/datasets/D:python新/建文件夹 (2)/hakusenninnshiki.v2i.yolov11/lane.yaml/images/val'
Note dataset download directory is '/content/datasets'. You can update this in '/root/.config/Ultralytics/settings.json'
it always show that

[glenn-jocher]

@sjocnjfjd the error indicates that the dataset path specified in your lane.yaml file is invalid or doesn't exist. Ensure the images/val folder is correctly located at /content/datasets/D:python新/建文件夹 (2)/hakusenninnshiki.v2i.yolov11/lane.yaml/images/val. Update the path in the lane.yaml file accordingly or modify your folder structure. For reference on dataset formatting, see Tips for Model Training.

[UltralyticsAssistant]

👋 Hello, thank you for your interest in Ultralytics 🚀! It seems like you're exploring the Ultralytics HUB Inference API, which is a great tool for integrating inference capabilities using Python or cURL 🌐. We recommend checking out the Docs for detailed usage instructions, examples, and guidelines. Many of the common queries are addressed there as well.

If this is a 🐛 Bug Report, please include a minimum reproducible example to help us debug effectively.

If you have ❓ Questions about custom implementations or API integrations, please share additional context, specific code snippets, or examples where you're facing challenges. This will help us understand your use case better and provide more accurate guidance.

Need real-time community support? Jump into our Discord 🎧. Want long-form discussions or to share your experiences? Visit Discourse or engage with enthusiasts on our Subreddit.

Upgrade

First, ensure you have the latest ultralytics package installed in a Python>=3.8 environment:

pip install -U ultralytics

This will help confirm if your issue persists in the latest version.

Environments

YOLO and Ultralytics tools can be run seamlessly in any of these verified environments with pre-installed dependencies:

• Notebooks with free GPU:
• Google Cloud Deep Learning VM (Quickstart)
• Amazon Deep Learning AMI (Quickstart)
• Docker Image. See our Docker Quickstart Guide

Status

The green badge above indicates that all Ultralytics CI tests are currently passing. These tests verify the correct operation of all YOLO Modes and Tasks on various platforms.

This is an automated response 🤖. An Ultralytics engineer will review your discussion and assist you soon. Thank you for your patience and for being part of the Ultralytics community! 😊

[sjocnjfjd]

Hi, thanks for the reply. I think there may be a misunderstanding. I am training on&nbsp;Ultralytics Hub, not on my local machine. My dataset path on the Hub is&nbsp;/datasets/a2O66UeTvAEqSUPPff6l/lane/. The path to my validation data in my&nbsp;lane.yaml&nbsp;file is&nbsp;images/val, which is the correct relative path to the&nbsp;lane.yaml&nbsp;file itself. My&nbsp;lane.yaml&nbsp;is located within the&nbsp;lane&nbsp;dataset directory on the Hub. The issue I'm experiencing is that after the model training completes, the Hub shows a "Disconnected" status and I cannot deploy the model. It's&nbsp;not&nbsp;a problem of the training process finding the dataset paths during training. The training itself completes without errors. The path you provided&nbsp;/content/datasets/D:python新/建文件夹 (2)/hakusenninnshiki.v2i.yolov11/lane.yaml/images/val&nbsp;is incorrect, as it mixes the Hub path&nbsp;/content/datasets/&nbsp;with my local Windows path&nbsp;D:python新/.... Local Windows paths are not accessible when training on the Hub. I've attached screenshots of my training code, the model page on the Hub, and the complete training logs. Could you please take another look?

…

---Original--- From: "Glenn ***@***.***&gt; Date: Sat, Dec 7, 2024 00:06 AM To: ***@***.***&gt;; Cc: ***@***.******@***.***&gt;; Subject: Re: [ultralytics/ultralytics] zh/hub/inference-api/ (Discussion#18075) @sjocnjfjd the error indicates that the dataset path specified in your lane.yaml file is invalid or doesn't exist. Ensure the images/val folder is correctly located at /content/datasets/D:python新/建文件夹 (2)/hakusenninnshiki.v2i.yolov11/lane.yaml/images/val. Update the path in the lane.yaml file accordingly or modify your folder structure. For reference on dataset formatting, see Tips for Model Training. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: ***@***.***&gt;

[glenn-jocher]

@sjocnjfjd thank you for the additional details and clarification. The "Disconnected" status after training typically indicates an issue within the Ultralytics HUB services. Please ensure that your project and model setups in HUB are complete and correctly linked. If the issue persists, I recommend raising a support ticket directly on HUB or reaching out via the GitHub HUB Issues page for assistance with the deployment issue.