Metrics for Azure Custom Vision and default value for Overlap Threshold when using Prediction API

Tom Jenkins 46 Reputation points
2021-01-05T09:04:06.273+00:00

Hello,
I'm using Azure Custom Vision for a research project.

I would like to know some things about the Model:

  • How does the model calculate the average precision?
  • When using the the prediction API, what are the settings for the Overlap Threshold?

(If i set the Overlap Threshold to 100 an send an Image via Prediction API to the Endpoint, i still get results, although the Precision and Recall are 0. That's why i think there must be a default value for the Overlap Threshold, that cannot be changed. Does that make sense? If yes, my question would be:
What is the default value for the Overlap Threshold, when posting via Predicition API?)

Thank you for the Support.

Azure AI services
Azure AI services
A group of Azure services, SDKs, and APIs designed to make apps more intelligent, engaging, and discoverable.
3,650 questions
{count} votes

Accepted answer
  1. GiftA-MSFT 11,176 Reputation points
    2021-01-05T17:25:34.617+00:00

    Hi, mean average precision is the average value of the average precision (AP). AP is the area under the precision/recall curve (precision plotted against recall for each prediction made). It seems you are referring to the Overlap Threshold for Get Iteration Performance API. If so, the default value if you didn't provide any is 0.5. However, from our portal we set the threshold to 0.3 when making a call. Hope this helps.

    1 person found this answer helpful.

0 additional answers

Sort by: Most helpful

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.