Custom Vision AI training has been stuck for hours

Hello Nadia Ardian,

Thank you for reaching out on the Microsoft Q&A.

If your Custom Vision training has been stuck for hours, it’s usually related to high training traffic or resource availability. During busy periods, many customers might be running training jobs at the same time, which can cause delays in the queue. While earlier iterations may have finished in minutes, later ones can take longer if the system is under heavy load.

It’s also important to note that once training has started, there’s currently no option to manually stop or cancel it. The system will eventually finish the job or stop automatically when your training budget is reached. You also won’t be billed beyond the set budget, so even if it looks like it’s running indefinitely, the charges remain capped.

If this continues to happen often, it’s worth double-checking your training budget and resource tier. For larger datasets or advanced training options, you may need to allocate more training hours. Quick Training is usually faster but depending on the image volume and complexity of the project, training time can vary.

In most cases, the best approach is to give it more time and let the system handle the queue. Once the training completes, you should see the results reflected in the iteration list.

Please find the attached document for your reference:

• Custom Vision frequently asked questions
• Quickstart: Create an object detection project with the Custom Vision client library (programming-language-python)

If you feel that your quires have been resolved, please accept the answer by clicking the "Upvote" and "Accept Answer" on the post.

Thank you!