Azure Custom Vision : Training Algorithm Details

Question

Hi,

We are using Azure custom vision to make object detection and classification models. We have trained our models for multiple hours and validated its accuracy. However, we don't have much information on details about the algorithms being used by Azure to train the model. Appreciate if below details can be provided :

1. Algorithm being used while training models - object detection and classification models
2. Details on deep learning algorithms being used for training models - object detection and classification models
3. Details on how many layers of DNN/algorithms is being used while training - object detection and classification models
4. What is the difference in algorithms/layers of DNN when we train the model for different hours. For example: What is the approach if I train model for 2 hours, 10 hours or 24 hours.
5. How many iterations are run for each training cycle.
6. How many epochs are used for training.
7. How Azure decides on selecting images from training datasets for testing purpose
8. How Azure calculates Average Precision and Mean Average Precision

We need details for our understanding and take a holistic approach toward design making decisions.

Regards,
Amit

Answer 1

@AmitKumar-7735 Few of the requested details can not be shared as they are confidential, Request you to raise azure support ticket to discuss offline.