How to monitor object detection model in production?

Hi Everyone, I have a question, can someone tell me how to monitor the object detection model in production, relevant articles would be helpful.

I have build FasterRCNN using PyTorch, I’m moving my model to production I would like to know which statistical and which parameter we consider to monitor the model or retrain the model.

Earlier, I have productionized regression model into production and I have used statistical test to monitor the data drift and model drift.

can someone tell me how to monitor the object detection model?

Thanks in advance :slight_smile:

related:

by the way: proper english would be “I have a question”.