Image resolution for inference

[Micky-123]

Hi,
What is the default image resolution on which the model operates? I tried with 360p and 1080p and it took the same time, so I guess the middleware code is rescaling all the images to a particular resolution before doing inference. Can you please let me know.
Also, is the default model based on resnet-50 or resnet-152?
I would like to experiment with different models. Is it possible to download models based on different resnet backbone and mobilenet backbone?
Thanks!

[serengil]

1- it does not have a pre-defined shape

# https://github.com/serengil/retinaface/blob/master/retinaface/model/retinaface_model.py#L55
data = Input(dtype=tf.float32, shape=(None, None, 3), name='data')

2- backbone is resnet-50 as mentioned here and here. no you cannot switch models.