Tensorflow Object Detection model 320x320 and 640x640 difference

Question

I am working on a project using TensorFlow API to perform object detection implement in Android app. In TF2 Dectection Model Zoo, there are same model with different feed in image size, such as SSD MobileNet V2 FPNLite 320x320 and SSD MobileNet V2 FPNLite 640x640. Some said that resize images by using fixed_shape_resizer and keep_aspect_ratio_resizer in config file.

I look through the pipeline.config file of SSD MobileNet V2 FPNLite 320x320 and SSD MobileNet V2 FPNLite 640x640, the only difference is that

SSD MobileNet V2 FPNLite 320x320

image_resizer {
  fixed_shape_resizer {
    height: 320
    width: 320
  }
}

SSD MobileNet V2 FPNLite 640x640

image_resizer {
  fixed_shape_resizer {
    height: 640
    width: 640
  }
}

Does it mean that SSD MobileNet V2 FPNLite 320x320 and SSD MobileNet V2 FPNLite 640x640 are actually the same if I change the image_resizer parameter?

My training images and detection are 2560x1440 (image size of Android mobile phone). Is that if I change the pipeline.config of either SSD MobileNet V2 FPNLite 320x320 or SSD MobileNet V2 FPNLite 640x640 to

image_resizer {
  fixed_shape_resizer {
    height: 2560
    width: 1440
  }
}

, then no resize will be done on image when feeding in the model? Thanks.