1

I have trained two Faster RCNN models with the exact same parameters with the exception of:

  1. Model 1: 150 epochs
  2. Model 2: 5 epochs

Once the models have been trained, I load them in and then I make one prediction on the same image for each model.

Inference prediction time:

  1. Model 1: 70 sec
  2. Model 2: 5 sec

To be clear, the time above is AFTER the model has been trained not during training.

Question: How can the inference time increase with the number of epochs?

Libraries:

  • tensorvision
  • torch
Ole V.V.
  • 81,772
  • 15
  • 137
  • 161
RaLa
  • 21
  • 2
  • Are you sure, you are doing everything correctly? It can't be such a difference within 2 models, trained on same dataset, with higher epoch nr.. Btw is it a torch or tensorflow implementation? I see you put tags with tf, then libraries w torch.. – TkrA Apr 29 '22 at 12:33
  • yea I have compared the state_dicts and they are exactly the same. Oh sorry its Torch. I am implementing it in Azure using AutoML. – RaLa Apr 29 '22 at 13:23

0 Answers0