Cognitive Services - Custom Vision Training Requirements

Lucas Lietz 21 Reputation points
2020-11-26T12:47:43.077+00:00

Hey there,

since I couldnt find any valid information about the requirements for custom vision object detection training, i was wondering if you could help.
What is the requirement for the training api. I have uploaded 30+ images in tag 1 and 5 images in tag 2.
I read that at least 30 images would be sufficient. On another source it says 15 per tag. Whats right here?

Thank you,
Lucas.

Azure AI Custom Vision
Azure AI Custom Vision
An Azure artificial intelligence service and end-to-end platform for applying computer vision to specific domains.
247 questions
Azure AI services
Azure AI services
A group of Azure services, SDKs, and APIs designed to make apps more intelligent, engaging, and discoverable.
2,833 questions
Accepted answer
  1. romungi-MSFT 45,961 Reputation points Microsoft Employee
    2020-11-27T10:20:53.227+00:00

    @Lucas Lietz I believe you are referring to the 30 images per tag as per the recommendation here. This can be the initial training set for training the model to detect your object but this could vary based on the image quality to a higher number or the minimum number 15 could be sufficient as mentioned in the limits page. 15 is the minimum for your project per tag and this can be any number higher than that based on the accuracy you achieve and no. of iterations used. We hope this clarifies the number details mentioned in these two pages and you could use a number that helps you give the required accuracy.

0 additional answers

Sort by: Most helpful
Most helpful Newest Oldest

Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.